4118 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO.

4, DECEMBER 2022

A Comprehensive Review on Fuel Cell UAV Key

Technologies: Propulsion System, Management
Strategy, and Design Procedure
Liangcai Xu , Member, IEEE, Yigeng Huangfu , Senior Member, IEEE, Rui Ma , Senior Member, IEEE,
Renyou Xie , Member, IEEE, Ziyou Song, Member, IEEE, Dongdong Zhao , Member, IEEE,
Yongheng Yang , Senior Member, IEEE, Yupeng Wang , Senior Member, IEEE, and Liangfei Xu

Abstract— Unmanned aerial vehicles (UAVs) can be regarded that UAVs can be used for dangerous and difficult missions [2],
as one of the most emerging technologies. Compared with piloted which will significantly guarantee the worker’s safety and
aerial vehicles, UAVs qualify for higher safety and lower cost, and health. Moreover, it is also convenient for humans to execute
they are more suitable for dangerous missions. Recently, fuel
cell (FC) UAVs have obtained more and more attention due to remote operations by UAVs [3]. Therefore, the efficiency of
their relatively higher energy density. However, their propulsion the whole human society is greatly improved. Meanwhile, due
system still needs to be further improved to meet different to the development of advanced manufacturers and advances
requirements of performing difficult tasks. This article aims to in material science, various types of UAVs can be found on
give a comprehensive review of the key technologies of FC UAVs. the market. For example, UAVs can be classified as fixed
First, various architectures of the propulsion system and corre-
sponding advantages and disadvantages are introduced. Then, wing, rotary wing, and flapping wings when their shapes
three kinds of management strategies for FC UAVs are reviewed. are considered [4]. Furthermore, regarding the power sup-
Next, some special applications and considered design procedures ply sources, there are solar-powered, battery-powered, fuel
are summarized. Compared with other review articles, this article cell (FC)-powered, and internal combustion engine (ICE)-
tries to cover more aspects of FC UAVs. Since these aspects powered UAVs. Moreover, respecting size, they can be catego-
have a tight relationship with each other, giving a comprehensive
review can be more beneficial to us to understand the state- rized as nano, micro, mini, medium, and large UAVs [5]. All
of-the-art development of propulsion systems for FC UAVs. these UAVs could be applied in various tasks, such as traffic
The conclusions show that advanced propulsion architectures, monitoring, infrastructure inspection, monitoring, environmen-
efficient online energy management strategies, and professional tal monitoring, delivery service, agriculture preservation, wire-
design procedures are still in urgent demand in the commercial less coverage, and even military applications [6]. According
process of FC UAVs.
to the market research report released in 2019, the global
Index Terms— Aviation, design procedure, energy manage- commercial UAV market size would reach U.S. $17 billion
ment strategies (EMSs), fuel cell (FC), propulsion architectures, in 2024 [7]. Because of these tremendous benefits, more and
unmanned aerial vehicle (UAV).
more companies and multinationals are putting a lot of effort
I. I NTRODUCTION into the improvement of UAVs.
As one of the most crucial parts, the innovations in propul-
U NMANNED aerial vehicles (UAVs) can be considered
flying robots, which have received much research efforts
from scientists in the past few decades [1]. The main reason is
sion systems have been the primary driver of the progress in
UAVs. In the last few decades, the ICE always dominated
as the main power source due to its high power and energy
Manuscript received 3 April 2022; revised 12 June 2022; accepted 19 July density [8]. However, the ICE-based UAVs also have many dis-
2022. Date of publication 1 August 2022; date of current version 21 October advantages, such as high noise, high cost, high vibration, and
2022. This work was supported in part by the National Natural Science
Foundation of China under Grant 52007155 and Grant 61873343 and in especially pollutants emissions (i.e., greenhouse gases, NOx
part by the Aero Science Foundation of China under Grant 20200019053005. and SO2 ). To reduce their impacts on the environment, hybrid
(Corresponding author: Rui Ma.) architecture for the propulsion system has been proposed and
Liangcai Xu is with the School of Automation, Northwestern Polytechnical
University, Xi’an 710072, China, and also with the Department of Mechanical researched deeply, which contains an ICE source and battery
Engineering, National University of Singapore, Singapore 117575. source [9]. To some extent, this method is effective. However,
Yigeng Huangfu, Rui Ma, Renyou Xie, and Dongdong Zhao are with the when considering the shortage of traditional foil fuel sources
School of Automation, Northwestern Polytechnical University, Xi’an 710072,
China (e-mail: [email protected]). and serious environmental pollution, this hybrid method is
Ziyou Song is with the Department of Mechanical Engineering, National not appropriate. Moreover, the investment in the international
University of Singapore, Singapore 117575. policy to phase out ICE-based vehicles has already shown that
Yongheng Yang is with the College of Electrical Engineering, Zhejiang
University, Hangzhou 310027, China. the electrification of aviation is a big trend [10].
Yupeng Wang is with the Powertrain Department, General Research and Until now, there are many kinds of electric power
Design Institute, FAW, Changchun 130011, China. sources, such as fuel cells, solar panels, batteries, and
Liangfei Xu is with the State Key Laboratory of Automotive Safety and
Energy, Tsinghua University, Beijing 100084, China. supercapacitors [11]. Among them, the battery can be said
Digital Object Identifier 10.1109/TTE.2022.3195272 to be the most widely used electric power source. How-
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4119

ever, it has relatively low energy density and long charger cell UAV demand. Therefore, this article will first try to give
time, which leads to UAV’s poor endurance and low efficient a relatively comprehensive review of some key technologies
flight time [12]. Up to date, the endurance of battery-powered of the whole propulsion system for fuel cell UAVs. From the
UAVs has already reached 90 min due to the development of authors’ point, since all the technologies of propulsion systems
lithium-polymer batteries [13]. If a longer endurance time is have a tight relationship with each other, a comprehensive
required, the quantity of batteries needs to be increased. Some- review will be more helpful to readers to understand them
times, this solution is not practical because of the constraints and be beneficial to the future development of fuel cell UAVs.
of weight and space for the UAVs. Therefore, an alternative The rest of this review article will be organized as fol-
method should be proposed. This method should not only lows. Section II presents the different architectures of electric
supply additional energy but also satisfy the limited conditions. propulsion systems for these UAVs. The EMSs for fuel cell
Solar panels and fuel cells are two preferred candidates. Solar UAVs are given in Section III. Some special applications and
panels have already been widely applied to fixed-wing UAVs. corresponding design methods for different fuel cell UAVs
In this way, solar energy could be converted to batteries during can be found in Section IV. Finally, some conclusions will
the daytime. Meanwhile, those solar panels almost do not be made in Section V.
affect the whole aircraft (i.e., space, weight, and aerodynamic
structure) because they can be tightly attached to wings. How-
II. E LECTRIC P ROPULSION A RCHITECTURES
ever, solar panels are only convenient for fixed-wing aircraft
FOR F UEL C ELL UAV S
and their performance could be significantly affected by the
weather. What is more, additional large batteries must be used The architecture and operation of the electric propulsion
to keep the UAVs frying during the night. By contrast, fuel system are one of the most important key technologies for
cells can not only be applied to any type of UAV but also fuel cell UAVs. The better system architecture will bring
can work continuously day and night. Besides, due to its better performance, such as longer endurance time and higher
propriety of high specific energy density, the endurance of fuel efficiency. Therefore, the objective of this section will focus on
cell-powered UAVs is relatively longer than the pure battery- sorting out the three most popular electric propulsion system
powered ones. architectures and summarizing their respective advantages and
It should be noted that fuel cell-powered UAVs have disadvantages. For convenience, only the fuel cell stack and
been researched and developed since 1959 [14]. Up to now, battery are selected as the power sources for the analysis of
although the performance of fuel cell UAVs has been signifi- the propulsion system.
cantly improved due to the contributions from a large number
of researchers, there is still a long way to go before commer-
cialization. Many technologies related to the fuel cell electric A. Pure Fuel Cell Power System
propulsion system need to be further improved to improve the Due to the various advantages of the fuel cell system,
durability, reliability, and fuel efficiency of fuel cell UAVs. Bradley et al. [24] developed a pure fuel cell-powered UAV in
Generally speaking, these key technologies mainly include 2007. The detailed fuel cell power plant diagram and the sys-
propulsion system architecture, energy management strategies tem specifications are shown in Fig. 1. This propulsion system
(EMSs), and design procedures for general fuel cell UAVs. For contains many important components, such as variable cathode
each key technology, there are several different selections or flow control, liquid cooling, self-humidification, and variable
solutions. Therefore, to help engineers design fuel cell UAVs cycle anode purging. To evaluate the overall performance of
that could meet commercial needs, a comprehensive summary the pure fuel cell-powered UAV, a comprehensive test flight
of these related technologies should be made. In this way, the was carried out. The flight test process includes idle speed,
potential challenges and prospects of these technologies can high power, and nominal cruise. Also, Figs. 2–5 show the
be summarized easily, which is very beneficial to the devel- propulsion system losses in detail under these three conditions.
opment of fuel cell UAVs. Recently, much review literature Under idle conditions, as shown in Fig. 2, when the
has been found to address different aspects of fuel cell-related 1.26 standard L/min hydrogen is supplied, a lower heating
technologies, for example, the strategies of the propulsion value (LHV) of 227 W can be obtained. However, almost
system on UAVs [15]; the fundamentals and applications of 74% of the energy is wasted due to the higher frequency
fuel cell technology [16]; evaluations of fuel cell stacks to anode purging, and the net electricity generated by the fuel
achieve high altitudes in light aircraft [17]; the conclusions cell system is almost zero. Therefore, the overall efficiency of
on barriers of scaling-up fuel cells [18]; proton exchange this power system is unsatisfactory. When the UAV started to
membrane fuel cell (PEMFC) system control methods [19]; take off or climb, it could be seen from Fig. 3 that the hydrogen
life cycle evaluation of hydrogen and other potential fuels utilization rate of the fuel cell propulsion system is above 88%.
for aircraft [20]; future aircraft design concepts for energy The efficiency of the motor controller and the propeller is
efficiency and environmental protection [21]; fuel cell APU 74% and 70%, respectively. Also, the efficiency of the whole
systems for aircraft [22]; and power supply and energy man- propulsion system is 14%. As for the cruise condition, shown
agement for aerial vehicles [23]. These articles mainly focus in Fig. 4, the UAV always maintains an altitude of 10 m
on one aspect, and most of them fail to analyze the current and maintains a stable airspeed of 13.6 m/s. Similarly, the
state of technology development from the perspective of fuel efficiency of the motor controller and the propeller is 66%
4120 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

Fig. 1. Detailed power plant diagram of the pure fuel cell-powered

UAVs [24].
Fig. 3. Propulsion system losses under high-power conditions [24].

Fig. 2. Propulsion system losses under the idle condition [24].

and 80%, respectively. The overall propulsion efficiency from

the hydrogen input to the propulsion system is about 18%.
The abovementioned research conducted by
Bradley et al. [24] has validated that the concept of
fuel cell aircraft is feasible. Besides, its operation complexity
Fig. 4. Propulsion system losses under cruise conditions [24].
and system weight are also satisfactory. However, low
efficiency, poor dynamic performance, and limited fault-
tolerant capability hinder the development of UAVs powered
by pure fuel cell systems. To make the pure fuel cell system
more feasible for fuel cell UAVs, a lot of efforts have been Fig. 5. Fuel cell and battery series hybrid electric propulsion configuration.
put in by other researchers, for example, advanced hydrogen
storage method [25], alternative onboard hydrogen generator
architectures can be summarized. One of them uses fuel cells
method [26], high power density fuel cell stacks that are
and batteries as power components, and the fuel cells are used
equipped with metal bipolar plates [27], high-efficiency
as the main power source. The other is to combine the fuel
power plant architecture [28], and a more compact balance
cell and the ICE to provide power for the UAVs, with the ICE
of plant (BoP) [29]. In addition, series hybrid and parallel
as the main power source. The characteristics and working
hybrid propulsion architectures are also common solutions to
principles of these two systems are discussed in the following.
improve the performance of fuel cell propulsion systems.
Fig. 5 shows a configuration diagram of a series of hybrid
propulsion systems with fuel cells and battery packs [30].
B. Series Hybrid Power System In fuel cell-powered UAVs, the fuel tanks, engines, and gener-
According to the published papers, patents, and other related ators can be replaced by fuel cell systems. Also, the fuel cell
documents, two main series-type electric propulsion system in the series configuration acts as an auxiliary power source to
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4121

air, which means that the reforming reaction process will

proceed successfully. The reformed products containing H2 ,
CO, CH4 , CO2 , and N2 flow into the anode channel of the
fuel cell, and then, electrochemical reactions occur in the fuel
cell in sequence. Fourth, part of the exhaust gas from the
anode channel of the fuel cell is recycled to the reformer, and
the rest flows into the combustor. In addition, some fuel is
also injected into the combustor to reach higher temperatures.
Finally, the combustion chamber exhaust gas is supplied to the
heat exchanger to heat the air entering the cathode channel of
the fuel cell. After that, the exhaust gas expands and outputs
propulsive force together with the nozzle. To further improve
the efficiency of the entire system, Aguiar et al. [36] conducted
a comprehensive study. A combined SOFC/gas turbine system
running on liquefied hydrogen is considered. The correspond-
Fig. 6. Fuel cell and ICE series hybrid electric propulsion configuration.
ing results show that, compared with the single-cell stack
(a) Scheme diagrams of the turbine-less jet engine integrated with a fuel cell design, when the fuel cell stack is modularized into multiple
and steam injection. (b) Possible actual configuration of the engine [35]. discrete stacks, the fuel is distributed in parallel, and the air
is supplied in series, the system efficiency and system weight
can be substantially improved. Adopting this technology, the
charge the battery. Generally, in a series configuration system, UAV with a liquefied hydrogen payload of approximately
the battery needs a larger size to meet the peak power demand. 780 kg can achieve a one-week mission. The relevant system
In contrast, since the fuel cell is only responsible for charging architecture and the corresponding layout diagram are shown
the battery, a smaller size can be selected. Moreover, the fuel in Fig. 7. However, the operation of this architecture is quite
cell stack can always work at the best efficiency and only complex and the onboard liquefied hydrogen storage is not
start working when the battery’s state of charge (SOC) reaches practically feasible.
the predetermined threshold. Compared with a pure fuel cell In summary, due to fuel recovery and heat reuse, the
power system, a hybrid power system in which a fuel cell efficiency of the IEC and fuel cell series hybrid system is
and a battery are connected in series can avoid the waste of relatively higher than that of the traditional pure fuel cell
hydrogen, and the system’s efficiency is also improved when propulsion system. In addition, the thermal efficiency, specific
the fuel cell UAV is working in an idle state [31]. Some thrust, and impulse of the hybrid power system also have
tests are carried out in [99] and the results show that the obvious advantages over traditional turbojet engines [37], [38].
extra weight for the series configuration is primarily due to However, due to the existence of IEC engines, greenhouse
the required generator and larger electric motor. Its operation gases and other harmful gases will still be produced.
complex is simple because the pure fuel cell system and the
dynamic performance, and fault-tolerant capability are also
greatly improved. Moreover, since the UAV propulsion motors C. Parallel Hybrid Power System
usually require higher speeds, series hybrid systems are not The parallel hybrid propulsion system is the most used
quite suitable [34] and the fuel cell and battery series hybrid and researched configuration in fuel cell UAVs. Generally,
electric propulsion configuration may be more suitable for low- a parallel hybrid propulsion system can be divided into two
speed, high-torque applications, such as buses [32] and aircraft configurations: active configuration with controlled dc–dc con-
tractors [33]. verters, and passive configuration, which is directly connected
Fig. 6 shows a schematic of an ICE (turbine-less jet engine) to the dc bus among all the power sources.
integrated with fuel cell and steam injection [35]. From Fig. 8 shows the three main types of passive structures [39].
Fig. 6(b), we can see that the solid oxide fuel cell (SOFC) Although they seem to be slightly different, the principles of
is connected in series with the turbine-less jet engines. This operation are quite different. For Fig. 8(a), the batteries are
architecture is defined as a hybrid system in which an ICE used as the main power source, which is similar to Fig. 4. The
and a fuel cell are connected in series. The detailed operation fuel cell system operates as a “range extender” system. When
steps will be described as follows. First, the atmosphere is the SOC of batteries reaches a predetermined minimum value,
compressed together with the intake air. The steam injection the switch will be turned on and the fuel cell system will start
and evaporative cooling occur in the intake pipe. Then, a com- generating electricity. Part of the electricity generated by the
pressor is used to compress the steam, which is usually driven fuel cell system is provided to the motor, and the rest is used
by an electric machine. The compressed flow is divided into to charge the lithium-ion battery. This process continues until
two parts: one part is supplied to the reformer and the other the SOC of the batteries reaches its predetermined maximum
part is supplied to the fuel cell cathode channel. Third, anode value. Since lithium-ion batteries play an important role in
exhaust gas, fuel, and water are mixed and compressed by the this structure, their selection and design are very important
blower to provide effective pressure for anode recirculation. to the performance of the entire system. Generally, to meet
The reformer is supplied with compressed fuel, water, and the high-power demand caused by the large load disturbance,
4122 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

Fig. 7. Different stacks of SOFC combined with GT. (a) UAV power system architecture. (b) One-stack SOFC/GT system arrangement diagram. (c) Two-stack
system arrangement diagram. (d) Three-stack system arrangement diagram [36].

Fig. 8. Passive hybrid fuel cell/batteries power system configuration [39]. (a) First type with one diode following fuel cell stack. (b) Second type with a
battery charger and two diodes following both fuel cell stack and battery. (c) Third type with two diodes following both fuel cell stack and battery.

the battery needs to have the ability of high capacity and high the operating principle, Fig. 8(c) is proposed. In this structure,
rated current, which in turn will lead to a larger system weight the fuel cell system and the battery are both connected with
or volume. In addition, the charging rate of the battery is not the dc bus through a diode. In [39], the propulsion system
limited or controlled, which may cause unnecessary damage integrates an open cathode, metal bipolar plates PEM fuel cell
to the battery. Therefore, a battery charger unit is added in stack, with fan, from the manufacturer BCH, model TH-200.
Fig. 8(b). Another difference between Fig. 8(a) and 8(b) is Meanwhile, six KOKAM Ultra High Energy NMC Lipo cells
that the fuel cell system is always connected to the dc bus are also adopted. The results show that to guarantee a suitable
through a diode. The fuel cell system and the lithium-ion joint operation of the fuel cell and the battery, careful design,
battery supply power to the motor together, and the dc bus selection, and development of both components are required.
voltage is clamped by the battery output voltage. According to However, compared with pure fuel cell systems and series
the polarization characteristic curve of the fuel cell, the output hybrid propulsion systems, both the dynamic performance and
terminal voltage of the fuel cell corresponds to its output fault-tolerant ability are improved.
current one-to-one. Under normal operating conditions, the So far, passive parallel hybrid systems can be seen in
output voltage of lithium-ion batteries changes within a narrow many research papers [40] and various commercial vehi-
range. Therefore, the fuel cell system can be prevented from cles, such as Honda Insight, Civic, and Accord hybrids [41].
operating under too low- or too high-power conditions. At the Verstraete et al. [42] conducted an experimental study to better
same time, the battery charger is responsible for controlling the understand passive power plants, including the importance of
charging current rate and charging time. Therefore, the battery battery and system energy management for proper system inte-
is also protected to extend its service life. To further simplify gration. The results showed that the battery played a vital role
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4123

structure B. The dc–dc converter is placed at the end of the bat-

tery, and the fuel cell system is directly connected in parallel
with the front-end dc–dc converter. Through this structure, the
discharge and charging of the battery can be well controlled.
Therefore, the battery life is prolonged and the current ripple
of the fuel cell is eliminated. However, the power response
of the fuel cell system is poor, which limits the dynamic
performance of the entire system. Moreover, when the hybrid
power system is applied to high-voltage operating conditions,
an additional dc–dc converter is required, which will cause an
additional cost. In structure D, both the fuel cell system and
the battery are connected through a dc–dc converter. With the
help of a dc–dc converter, the power sharing between these
Fig. 9. Ground test bench of the passive parallel hybrid system [43].
two power sources can be precisely controlled. The discharge
and charging of the battery are also well-controlled, which can
in the dynamic load response, and the dynamic performance ensure that the battery is always working in an appropriate
greatly depended on the battery’s capacity and rated current. state. Moreover, in this structure, the selection of the fuel
When the load changes rapidly, the performance of a battery cell system and the battery is more convenient than that of
with a small capacity but a large rated current is similar to that the other three structures. However, since the battery is not
of a battery with a large capacity but a small rated current. directly connected to the dc bus, the dynamic performance of
This finding can be regarded as design guidance for application the entire system may be degraded to a certain extent. To solve
engineers. Some other researchers also use ground test benches this problem, appropriate EMSs and corresponding control
to evaluate the performance of passive parallel hybrid power algorithms need to be carefully designed. Unfortunately, the
systems [43], [44], including the static behavior, dynamic cost of the entire system will become higher and the reliability
behavior, and battery charging characteristics of the entire will degrade. In order to validate the abovementioned active
system. The related picture of the ground test bench is shown architectures, a 1.5-kW prototype is built with a 1.2-kW fuel
in Fig. 9. The results verified that the passive configuration cell and 36-V 105-Ah lead-acid battery. Taking architecture D
could achieve good system efficiency and good response to as an example, due to the dc–dc converters, it can emulate
dynamic loads and could charge the onboard battery through all the working modes, which are similar to another three
a simple circuit without a dc–dc converter. architectures. The results show that this architecture maintains
In addition to the passive parallel hybrid power system, the the best flexibility and the highest efficiency. However, its
active parallel hybrid power system has also received a lot of operation complexity is unsatisfactory. Moreover, compared
research attention. Fig. 10 gives a comprehensive description with the passive hybrid propulsion system, when they both
of the various active structures [45]. In this figure, one of have the same power sources and loads, the total system
the black boxes represents the fuel cell system, and the other weight and fault-tolerant ability of the active hybrid propulsion
represents the battery. In structure A, the battery is directly system are both unsatisfactory. Fig. 11 gives a clear spider
connected to the fuel cell system, and then, the dc–dc converter diagram to evaluate the advantages and disadvantages of the
is placed between the dc bus and the load (i.e., the dc–ac above four active structures.
converter in the figure). The battery pack as an energy buffer In summary, the above two configurations have their advan-
is responsible for transient loads. At the same time, it also tages and disadvantages. Due to the control and management
functions as part of the load in a steady state. The output strategies of related converters, the active configuration allows
power of the fuel cell system is not controlled. Whenever the the selection and operation of fuel cells and batteries to be
output voltage of the battery is lower than that of the fuel decoupled. In addition, due to the well-controlled dc–dc con-
cell, the battery is charged. In this case, the dc–dc converter verters, each power supply can be precisely controlled, and
needs to handle all the power from two sources, which leads the power sharing between different power supplies is more
to an intensive current design. In structure B, the fuel cell appropriate. This means that both the fuel cell and the battery
system is connected to the battery through a dc–dc converter, can work under optimal conditions. Therefore, the lifetime of
which means that the high-voltage battery needs to match the entire system may be extended to its theoretical maximum.
the dc bus voltage. There are many similar characteristics However, it still needs to be noted that these well-controlled
between structures A and B. For example, the batteries in both converters also bring many shortcomings, such as low reli-
structures are directly coupled to the dc bus, which enables the ability due to more complex system topology and control;
power system to respond quickly to load transients. However, reduced system efficiency due to multilevel energy conversion;
this also means that the process of discharging and charging and larger volume and weight.
the battery is uncontrollable. Therefore, the battery may be For the passive configuration, it has the advantages of
overcharged or overdischarged, thereby accelerating battery higher efficiency, lower cost, and simple architecture. How-
aging. Moreover, since the dc bus voltage is determined by ever, it cannot achieve active power distribution and control
the battery output voltage, it is difficult to select a suitable among multiple power sources, which may cause the fuel cell
battery pack for UAVs. As for structure C, it is the opposite of or battery pack to be in poor working condition. On the other
4124 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

Fig. 10. Different active parallel hybrid power system structures [45].

advantages, such as high power density, high energy density,

fast response, high reliability, and long lifetime. Therefore,
when different types of power supplies are used in a parallel
structure, this structure can combine their respective excellent
characteristics to meet various load requirements, especially
under severe conditions. In addition, the parallel configuration
also has the best fault tolerance. Table II gives a brief com-
parison of pure, series hybrid, and parallel hybrid systems,
focusing on complexity, efficiency, fault tolerance, dynamic
performance, and weight. From this table, it can be found that
the parallel configuration will be more beneficial to fuel cell
UAVs in future development.

III. EMS S FOR F UEL C ELL UAV S

As mentioned in Section II, the fuel cells are always hybrid
with other power sources to complete a flight due to their
slow response. Verstraete et al. [42] tested a fuel cell/battery
propulsion system for UAV and indicated that the battery plays
Fig. 11. Spider plot for evaluation of four different active parallel hybrid a crucial role in the response to dynamic load change and
structures.
the system response is strongly influenced by the battery’s
hand, to meet the performance required under specific tasks, capacity and current rating. The hybridization of different
fuel cells and battery packs need to be carefully designed power sources brings power allocation problems; thus, EMSs
and selected. When choosing the battery packs, a tradeoff must are desired. According to [46]–[48], the goal of the fuel cell
be made between the battery capability, boost-power availabil- hybrid system EMS can be threefold: first, to guarantee that
ity, battery mass, and mission endurance. Therefore, the entire the fuel cell can supply stable or optimal power to the load;
design process may become more complicated. Table I gives second, to exploit the advantages of different power sources
a detailed and comprehensive comparison between the active regarding their working characteristic, hybrid construction, and
configuration and the passive configuration [39]. flight factor; and third, to achieve a comprehensive energy
management optimization to prolong the lifespan of the power
sources and improve the system durability, stability, and over-
D. Discussion and Future Prospect all efficiency.
In general, the pure fuel cell power propulsion system is The modeling methods are quite important for the design
the simplest, but it cannot adapt to various load requirements. of EMS for fuel cell UAVs. According to the abovementioned
This means that the UAV cannot obtain the best performance, goals of the fuel cell hybrid system’s EMS, the power alloca-
and the lifetime of the fuel cell system will drop sharply. As for tion model of the hybrid system is the basic requirement [101].
the series configuration, it is slightly more complicated than However, this model ignores the specific characteristics of
the pure fuel cell system but has a better performance. Unfor- components, such as fuel cell systems, batteries, and power
tunately, due to the characteristics of the two-stage conversion, electronics. Therefore, it is not easy to propose a satisfactory
its energy conversion efficiency is low. In contrast, the parallel EMS method only based on this model. To synchronously
configuration may perform best. Different sources have their obtain multiple objections, for example, high energy efficiency,
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4125

TABLE I
C OMPARISON B ETWEEN A CTIVE AND PASSIVE F UEL C ELL /BATTERY H YBRID P OWER S YSTEM [72]

TABLE II
C OMPARISONS A MONG P URE , S ERIES H YBRID , AND PARALLEL P OWER S YSTEMS

high economical efficiency, and long lifespan of fuel cell stack A. Rule-Based EMS
and battery, the specific models should be given and embedded
into the framework of the EMS method. In this article, let us Rule-based methods, such as state machines and fuzzy
take the specific models of the fuel cell stack as an example, logic control, are widely used in EMS due to their simplic-
which can be divided into two categories, physics-based and ity, reliability, and computation efficiency advantages. State
data-driven models [102]. Since the fuel cell stack is a typical machine EMS divides the system into different states accord-
multiphase system, many specific models are proposed and ing to the load power and battery’s SOC with corresponding
embedded into the framework of the EMS method, for exam- control actions, which is the most used rule-based method.
ple, the efficiency model [103] and the lifetime degradation In [50], a rule-based method is proposed for a low-speed long-
model [105], [107]. Compared with these specific models, endurance UAV powered by fuel cell, photovoltaic (PV), and
the multiphase model can present more details of the fuel battery. Six sectors are identified based on the required power
cell stack [106]. With the help of an advanced computation and SOC of the battery, which can ensure flight endurance and
platform and efficient modeling method, multiphase models maintain an appropriate SOC. Similarly, a state machine con-
could be a potential research trend in the future. trol, which divides the control area into five states, for fuel cell
Furthermore, according to the power system structure, the UAV is developed in [51]. A rapid control prototype is built
EMS can be divided into passive and active EMS. For the to testify simulation result, and the overall efficiency is more
passive power system shown in Fig. 8, there is no dc–dc than 90% under different initial SOC states with the proposed
converter to control the power flow, and the power delivered method. To ensure that the energy utility rate difference of the
by the fuel cell and the battery is determined by the load fuel cell and the battery are controlled in an acceptable range
power, voltage of the fuel cell, and the battery. The passive when the fuel consumption is the least, a rule-based dynamic
EMS has the advantages of simplicity, weight reduction, and balance EMS is proposed [52]. Three states are set based on
power loss reduction associated with the power converter, the difference between SOC and the remaining hydrogen rate.
and it has been applied in fuel cell UAVs. For example, the Under the proposed scheme, the reliability of the power system
passive EMS is applied to a tail-sitter hybrid lift UAV, which is improved as no power sources will be depleted in advance.
is powered by the fuel cell and lithium battery [49]. Although Moreover, simulations and experimental tests are carried out
passive EMS is easy to be implemented, it cannot utilize the to testify to the proposed method. Results validated that the
different characteristics of the power sources, which results method can ensure good performance under different load
in low efficiency. Also, the passive EMS cannot maintain the profiles. In [53], a power and energy management system for
minimum SOC of the lithium battery, which will accelerate fuel cell UAV is developed, and the state machine is applied
the degradation of the battery. In this article, we will focus on for energy management. The system operation is divided into
active EMS. Generally, the EMS methods can be categorized three states: start-up state, charging state, and high power state
into rule-, optimization-, and intelligent-based methods, which based on the battery terminal voltage and dc bus current.
are shown in Fig. 12. Besides, the EMS is coupled with the fuel cell compressor
4126 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

Fig. 12. Detailed structures of some commonly used EMS methods for fuel cell UAVs. (a) Fuel cell UAV EMS overview. (b) State machine method. (c) Fuzzy
logic control. (d) MPC. (e) NN. (f) RL.

optimal power control, which achieved 2.6% energy saving operating in a high-efficiency area with an affordable power
compared with the constant compressor power. slope [56]. Simulations are implemented under NEDC, UUDS,
Unlike the precisely deterministic method such as state and user-defined driving conditions. Compared with the deter-
machine, the fuzzy logic control uses the approximate way ministic method, the proposed method gets higher efficiency.
rather than the precise way, which is robust to system uncer- To get a better fuel economy, an optimized fuzzy logic con-
tainties. To realize online EMS, an online fuzzy logic strategy trol for a fuel cell/battery hybrid vehicle is proposed [57],
for a fuel cell/battery hybrid powered UAV is proposed [54]. and the genetic algorithm is used to find the optimal para-
The pulsed-power profile is implemented to evaluate the effec- meters of the controller. Simulation results indicate that the
tiveness and real-time feature of the proposed method. Besides, hydrogen consumption of the optimized fuzzy logic con-
comparisons are made with the passive, state machine control troller has been reduced by 22% from the conventional fuzzy
methods under different flight missions. Experiment results logic controller. To ensure a longer endurance, a fuzzy state
indicate that the online fuzzy strategy gets the best perfor- machine (FSM) strategy is proposed for PV/FC/battery hybrid
mance considering the fuel consumption. Aiming at the high- electric UAV [58], where the fuzzy logic control deals with
efficiency operation of fuel cells, a fuzzy logic controller the power split between the fuel cell and battery and the state
for a 1.2-kW fuel cell/battery hybrid system is developed machine deal with the power flow of PV and battery. To ensure
in [55]. A 40-min flight scenario that includes five flight long endurance, the fuel cell gets the low priority to supply
phases is implemented in the simulation and hardware-in-the- power to the system. Under the FSM strategy, 26.7% of the
loop (HIL) experiment. Similarly, the fuzzy logic control is hydrogen fuel is saved compared with constrained thermostat
applied in a fuel cell/battery hybrid system to keep the fuel cell control (CTC).
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4127

Besides the state machine and fuzzy logic, there are also B. Optimization-Based Methods
other rule-based methods, such as switching technique, if– Different from the rule-based method, the optimization-
then rules, and frequency decoupling. To prolong the flight based methods can be designed to find the optimal result
endurance, a switching technique for a fuel cell/solar cell the designer wants to achieve, for example: to consume less
(with a battery) hybrid UAV is proposed in [47]. To meet hydrogen and assure a long lifespan. Based on the system
the load variation during the flight, the two power sources power flow and constraints, the objection function will be
are controlled by a solid-state relay to decide which one or built and the optimization method will be deployed to find
both output power. Such a mechanism is tested throughout a the optimal solution.
1.5-h flight, and the system stability is guaranteed. Similarly, Generally, the optimization-based EMS can be categorized
switching control logic is designed in [59] for a fuel cell/ into offline (global) and online (local) methods. The offline
PV hybrid system. In [60], a fuel cell/LiPo battery hybrid EMS, such as dynamic programming, linear programming, and
system is built to evaluate the performance of three archi- quadratic programming [56], utilizes the known load profile to
tecture models. If–then rule-based management is designed, provide a globally optimal solution. However, the load profile
which takes the battery operating voltage, fuel cell current, in real missions is unknown in most cases, and thus, these
state of the fill, and charge rates in the hydrogen cylinder and methods can only be applied offline and are often set as a
battery SOC into account. Both the ground and flight tests are benchmark to evaluate other methods. Dynamic programming
carried out to select the hybridization scenario and the battery. is used in [65] for a fuel cell/battery hybrid UAV to deter-
Based on the fact that the fuel cell’s dynamic responsibility is mine the optimal power flow that guarantees a long-endurance
slower than battery and supercapacitor, researchers proposed flight given the flight path. Four test conditions, steady level,
frequency decoupling methods for hybrid EMS. Specifically, turbulent level flight, cyclic power mission, and high power
the fuel cell responds to the low-frequency power, whereas climb, are used to test the EMS. Simulation results indicate
the battery/supercapacitor will deal with the high-frequency that the hybridization can decouple the aircraft constraint from
power demand [61]. Under such a method, the SOC of the the aircraft energy requirements, which can be beneficial to
battery will keep in a narrow range since the mean energy of improve the performance through the proper integrated design
the battery is close to zero. In [62], two frequency decou- process.
pling methods: Wiener filter and wavelet transform for a Pontryagin’s minimum principle (PMP) as one of the online
fuel cell/ultracapacitor hybrid UAV are tested. The simulation EMS methods tries to find a globally optimal solution by
results on a 1-kW level system indicate that the Wiener filter transforming the global optimization problem into a series of
gets a better performance considering the mean high-frequency minimal values and solving one minimal value each time [65].
component, efficiency, and mean SOC value. In [63], a fre- To realize online optimization, an adaptive EMS based on
quency separation rule-based for a fuel cell/battery-powered PMP for fuel cell/battery hybrid vehicles is proposed in [66].
quadcopter is proposed, which enables the energy sources to To realize online driving cycle prediction, the PSO-SVM-
be close to their nominal operating points and thus increases based driving pattern classifier is first built and an improved
their service life. Besides, the SOC of the battery keeps in an Markov-based velocity predictor is applied to predict the
acceptable range compared with the rule-based strategy that driving behavior under different driven patterns. Simulation
lets the fuel cell output a constant power, and 3% hydrogen results show that 4% hydrogen is saved compared with the
saving is achieved by an equivalent consumption minimization rule-based method, and the fuel cell average power change rate
strategy (ECMS). In that reference, the flight data are used for is relatively low, which is beneficial to its durability. Similarly,
simulation verification. a real-time PMP EMS for fuel cell buses is developed [67],
To investigate the different EMS’s effects, a compar- and over 20% hydrogen saving is achieved compared with the
ative study of EMS for a fuel cell hybrid emergency rule-based method. It should be noted that the real-time PMP
power system of more electric aircraft is given in [64]. is realized by a DBSCAN clustering method that matches the
Six EMSs, including state machine control, rule-based fuzzy current driving segment based on the history data. Neverthe-
logic, classical proportional–integral (PI) control, frequency less, the proposed method can only work for applications that
decoupling, fuzzy logic, and ECMS, are compared based have a fixed driving cycle such as buses and trains. Aiming
on simulations and experiments carried out on a 14-kW at optimizing the fuel economy and durability, an EMS based
fuel cell/battery/ultracapacitor hybrid system. The hydrogen on PMP satisfactory optimization for fuel cell hybrid systems
consumption, overall system efficiency, scope of the bat- is proposed [68]. The experimental result indicates that the
tery/supercapacitor SOC, and life cycle are compared. Results proposed method can significantly reduce the fuel cell output
indicate that the state machine control provides slightly better power fluctuation while preserving the similar hydrogen con-
efficiency and stress on the battery and ultracapacitor. The sumption to the standard PMP method. To test the character
frequency decoupling and fuzzy logic achieved the lowest of the existing EMS methods, the fuzzy logic, DP, PMP, and
fuel cell stress and the classical PI control consumes the improved PMP for a fixed-wing furl cell UAV through simu-
least hydrogen. Although the rule-based methods are simple lation based on a given load profile are compared [69]. It is
and reliable, experiences are required to design an acceptable found that the fuzzy logic is robust, but with higher hydrogen
EMS, and the optimal solution can hardly be reached. Thus, consumption, the improved PMP maintains the lowest fuel cell
optimization-based methods are proposed to ensure the system output power rate and saves 5.4% hydrogen compared with
working at its optimal points. fuzzy logic.
4128 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

In conclusion, both the DP and PMP can get a global proposed method is the potential to be implemented online.
optimal solution for the system, whereas the requirement To avoid the start–stop and rapid load change of fuel cells,
for the driving cycle information makes them only suitable an NN optimized by the genetic algorithm is proposed for
for the systems with a fixed route. Thus, local optimization the fuel cell hybrid vehicle [75]. Through the optimization
methods, such as ECMS and model predictive control (MPC), of the genetic algorithm, the NN can be trained pertinently, and
are proposed. The ECMS is an instantaneous optimization the trained network can consciously avoid specific outputs
strategy that calculates the control variable by minimizing the according to the requirement. Simulation shows that the
equivalent hydrogen consumption. start–stop of fuel cells has been reduced compared with NN
Generally, the ECMS uses a predetermined factor to esti- and ECMS. Besides, the fuel consumption is close to the DP
mate equivalent consumptions online, which cannot guarantee result. In the framework of RL, the RL agent receives the
the SOC in the acceptable range. To address this problem, current state St and reward Rt from the environment and then
the adaptive-ECMS has been proposed. For the sake of safe determines a proper action At to function on the environment,
landing for aircraft, a fuzzy logic-based ECMS (F-ECMS) and then, the environment will feedback St+1 and Rt+1 to
method is proposed to improve the power system efficiency the agent. The process [which is shown in Fig. 12(f)] will
and keep the SOC of the battery in an appropriate range [70]. be continued until the agent finds a proper strategy according
Two test cases are simulated, and the comparisons are made to the maximum accumulative reward.
with conventional ECMS and adaptive ECMS. Results show To realize real-time application and approximate global
that the F-ECMS can balance fuel consumption and SOC. optimization, an EMS based on K-nearest neighbor (KNN)
To decrease the hydrogen consumption and increase the dura- and RL is proposed for fuel cell hybrid vehicles. The KNN
bility of the power sources, an ECMS for fuel cell/ battery is used to predict the long-term average speed of the future
and ultracapacitor hybrid system is developed [71]. Under the driving segment, whereas the RL is responsible for the global
designed controller, the battery is worked as the main source, optimal approximation [76]. Simulation results indicate that
the fuel cell is operated to find the maximum efficiency point 6.14% of fuel can be saved and the start–stop times of fuel
and the ultracapacitor is dedicated to providing the peak power. cells are reduced by 21.7% compared with the rule-based
Simulation results indicate that the proposed method achieves method. In [77], an RL-based algorithm using an ECMS
a smooth fuel cell current and saves 2.16% hydrogen compared for a fuel cell/ battery, ultracapacitor hybrid vehicle is pro-
with the rule-based method. In [72], an adaptive multimode posed. The simulation result under different driving cycles
MPC for fuel cell hybrid vehicles to reduce fuel consumption indicates that the equivalent fuel consumption of the RL-
and improve fuel cell durability. Driving segments obtained by based method is less than that of ECMS. Moreover, the com-
the Markov pattern recognizer are taken into consideration to putation efficiency of the RL-based method is higher than
tune the controller. Simulation results under different driving that of the DP method. To improve the overall efficiency of
scenarios show that 87% of fuel cell power transient is reduced a fuel cell/battery hybrid vehicle, an improved Q-learning-
and 2% of hydrogen is saved compared with the single-mode based method that takes the power source and converter’s
benchmark strategy. To test the influence of the velocity fore- efficiency into consideration is proposed [78]. HIL experiment
casting method, three methods, including generalized exponen- and on-road test show that the overall efficiency reaches 52%
tially varying, Markov chain, and neural networks (NNs) of under the proposed strategy, which gains 8% higher than that
velocity predictors for MPC-based EMS of a hybrid electrical of the state machine. To ensure long endurance, an improved
vehicle, are compared [73]. It is found that the NN predictor double Q learning for the energy management of Hybrid UAVs
can get the highest accuracy. Moreover, with the prediction is proposed [79], [100]. Plenty of flight data is used to train the
results of NN, the MPC can save 7% more hydrogen than model, and the results indicate that the method has stronger
the ECMS. It should be noticed that the performance of MPC robustness and convergence speed. Besides, comparisons are
depends on the power reference provided in each prediction made with the dynamic programming and rule-based methods.
result, and thus, precision prediction algorithms are needed to Results indicate that when the RL method has been well
guarantee its performance. trained offline, it can be implemented in real time without
sacrificing many functions in fuel and economical saving com-
pared with the DP. Compared with the rule-based method, 20%
C. Intelligent-Based Methods of fuel saving is achieved. Although the intelligent methods
For the optimization-based method, there is a tradeoff have the advantages of being model-free, the high requirement
between precision and computation burden. To address this of data and high computation burden make it hard for online
dilemma, intelligent-based methods, such as NN and reinforce- optimization. Until now, there is no such method that has been
ment learning (RL), have drawn research attention recently. applied in the fuel cell UAV online optimization.
To improve service durability and fuel economy, a multilayer
perceptron is adopted for driving pattern recognition, and
then, a fuzzy energy management controller is used for power D. Discussion and Future Prospect
allocation for a fuel cell/ultracapacitor system [74]. More- To better illustrate the EMS for fuel cell UAVs, their fea-
over, a genetic algorithm is utilized to optimize the critical tures are listed in Table III. Currently, rule-based methods are
parameter, thus prolonging the fuel cell lifetime and reducing mostly used in industry due to their simplicity and reliabil-
hydrogen consumption. Simulation results indicate that the ity. Although the optimization- and intelligent-based methods
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4129

TABLE III
S UMMARY TABLE OF EMS FOR F UEL C ELL UAV
4130 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

TABLE III
(CONTINUED.) S UMMARY TABLE OF EMS FOR F UEL C ELL UAV

can ensure a longer flight and durability, more computational While most intelligent-based methods have many extreme val-
resources are needed; otherwise, the online EMS may some- ues and they can also be used to get satisfactory results without
times not be guaranteed. The main difference between the rule- physical models. As for the real-time applications, most rule-
based method and the optimization-based method is that the based ESM methods could be used to manage the whole
optimization-based method is more dependent on the accuracy system’s power flow online, whereas most optimization- and
of modeling. If the model cannot be provided, it is difficult to intelligent-based EMS methods could not be implemented in
implement the optimization-based methods, not to mention to real time due to the heavy computation burden. To help these
get satisfactory results. However, if the system model can be two kinds of methods be implemented online in the future,
obtained, the global optimization results can be easily gotten more efficient algorithms and advanced computation platforms
when compared with the rule-based methods. Furthermore, should be improved. Besides, it can be seen from Table III that
compared with intelligent methods, most optimization methods most of the existing methods are verified by simulations or
can be classified into the convex optimization group, which HIL, but few researchers have applied their advanced EMS to
means that only one maximum or minimum value exists. the real fuel cell UAV. There is a gap between the simulation
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4131

Fig. 13. Potential applications of fuel cell UAVs.

and real application as the real flight is far more complicated. of crops and estimate the crop yield. UAVs can also be used
Besides, advanced methods such as RL are emerging in the for crop dusting and other tasks, such as weed detection,
application of fuel cell hybrid vehicles, whereas they have tillage analysis, and tillage mapping [81]. In urban service
been rarely applied to fuel cell UAVs due to the complicated applications, UAVs play an important role in traffic monitor-
working condition and safety issues. Therefore, more ground ing. Compared with traditional methods, UVA is more flexible
or real flight tests with advanced EMS should be investigated and more accurate and has a larger coverage area [82]. The
to prompt the real implementation of EMS on UAVs. Regard- traffic information obtained by UAVs will help people choose
ing the power sources, most of the power system structures more suitable traffic routes. At the same time, because UAVs
consist of fuel cells and batteries, whereas the performance can obtain traffic information in real time, map data will also
of the battery may deteriorate at high altitudes, which is the be updated in time to provide people with more accurate
case that high-altitude UAVs may encounter. Thus, it would navigation information. In addition, some companies have
be a good choice to install a supercapacitor to improve the used UVA for fast package delivery [4] or medical delivery,
performance in high altitude and cold weather since the such as Google, Amazon, and JD. In terms of infrastructure,
energy stored in an ultracapacitor can be maintained at low many facilities require regular maintenance, which requires
temperatures. a large number of employees and machines to be deployed
in a large work area. Today, UAVs can perform autonomous
IV. S PECIAL A PPLICATIONS AND inspections on oil pipelines, signal towers, high-rise buildings,
C ONSIDERED D ESIGN M ETHODS bridges, highways, dams, and so on [83], [84]. In terms of
Due to the development of academic research and industrial entertainment, DJI is one of the most famous companies,
manufacturing, fuel cell UAVs can be used in almost all which allows people to take pictures in the air through UAVs.
businesses, such as precision agriculture, delivery services, At the same time, many professional teams are conducting
wildlife surveys, research, and rescue. As shown in Fig. 13, aerobatics and light shows through UAVs. In the military field,
this briefly lists the potential applications of fuel cell UAVs. the application of UAVs is more common and mature than
The application of UAVs in various fields is described in detail that in the civilian field. In reality, UAV technology was first
next. developed for military activities, which can greatly protect the
In agriculture applications, people prefer to use UAVs to lives of soldiers. The famous military applications of UAVs
capture high-precision images above the fields [80]. After include artillery guidance, reconnaissance flights, communi-
that, various image processing algorithms are used to obtain cation jammers, electronic warfare, antiship missile defense,
specific data. Those data can indicate the health information and so on [85]. In the power field, more and more renewable
4132 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

First, the proposed design process focuses on a predesigned

fuel cell power plant, as shown in Fig. 14(a) [93]. This is
the most convenient method because the power plant can be
considered an integral component. After the fuel cell power
plant is selected, it should be combined with the parametric
aircraft model. However, because the characteristics of the
power plant are derived from the automotive fuel cell, the fuel
cell subsystem or component-level model cannot be obtained.
Therefore, the disadvantage is that the interaction between the
fuel cell components cannot be modeled in detail, and the low-
level information about the power plant cannot be provided for
the implementation of subsequent tasks.
The second design method refers to the design of the fuel
cell and its subsystems are independent of the design of
aircraft [94]. Its configuration is shown in Fig. 14(b). In this
way, the fuel cell power plant can achieve the required specific
energy, and the interaction of the components in the subsystem
can also be estimated clearly. The disadvantage of this method
is that it does not consider the interaction between the fuel cell
and the aircraft.
The last method is the integration of power plant and
Fig. 14. Three design processes were used in the previous study [93]. aircraft [95]. As shown in Fig. 14(c), the aircraft level and
(a) Configuration A. (b) Configuration B. (c) Configuration C. the fuel cell power plant level are performed simultaneously.
In this way, the aircraft and fuel cell power plants can be
energy power generation equipment is integrated into the grid, optimized at the same time, which means that the final power
such as solar panel power stations and wind power plants. plants will be more suitable for specific applications. However,
UAVs have been used to estimate the dust accumulation on it requires more design space, which will further lead to greater
solar panels [86] and detect the fault of the wind turbine blade complexity and higher computational costs.
surface [87]. With the maturity of high-voltage transmission Through the review of the above three design processes,
technology, UAVs are also used for the inspection of power it can be found that the last one may be more competitive.
lines [88]. Natural disasters also pose a threat to people’s lives To solve its disadvantages of high complexity and high com-
and properties. Therefore, it is very important to use UAVs to putational cost, many researchers have conducted a lot of
perform regular monitoring and data collection in forests, vol- research on it. Next, three specific design procedures for fuel
canoes, and even the atmosphere to help us develop effective cell UAVs will be introduced in detail.
prevention strategies [89]. Moreover, when unexpected disas-
ters occur, the communication infrastructure may be destroyed.
In this case, the UAV can become a signal relay station and A. Specific Design for High-Altitude Fuel Cell UAV
quickly provide data connections [5]. If the damaged area is For a high-altitude fuel cell UAV, the desired altitude is
large, UAVs equipped with infrared cameras can also be used usually above 10 km, which means that it will take longer
to detect life signals. In the field of environmental protection to climb. Therefore, generally speaking, the fuel cell system
and scientific research, UAVs are also regarded as an effective needs to be used as the main power source, and the battery
tool. When a specific sensor is installed on the UAV, it can is only used for starting and powering the control board. The
perform real-time monitoring of pollutant gas diffusion rate design process of the high-altitude fuel cell UAV is given in
[90], water pollution [91], and biological migration. Govern- the flowchart, as shown in Fig. 15.
ments around the world are paying more and more attention to First, it is necessary to obtain or determine the main char-
these activities. In addition, when working in the field, UAVs acteristics of the UAV, such as the total mass of the fuselage,
can be used for data collection in hazardous environments. the expected service ceiling, the lift coefficient, and the wing
Through the abovementioned various UAV application cases load. Then, simple aerodynamic principles need to be used to
and the various advantages of fuel cell UAVs, it can be seen calculate the minimum power at the predesigned service ceil-
that the demand for fuel cell UAVs will increase greatly. ing. Third, the power conversion efficiency of each component
To better satisfy this demand, a mature design process for fuel should be estimated empirically. Through all these efficiency
cell UAVs is important. The rest of this section will focus values, the total efficiency of the power system can also be
on the general design procedures for most fuel cell UAVs obtained. Applying this total efficiency value to the equation
and detailed design methods for specific applications. In the shown in the first stage, the relationship between the mass
early days, there are few reports about fuel cell UAV design of the UAV and the minimum fuel cell power is obtained.
methods; therefore, the design methods for conventional air- This equation gives the upper limit of the total mass of the
planes are usually made as a reference. At present, three design UAV. Next, the fuel cell stack system and its BoP should be
methods are mainly used [92]. carefully selected. In this step, the rated power of the fuel cell
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4133

Fig. 17. Conceptual design process of small fuel cell UAV [97].

Fig. 15. Design flowchart of high-altitude fuel cell UAV [19].

Fig. 18. Design flowchart of long-endurance fuel cell UAV [97].

attractive than the former. If the latter is selected, we can also

calculate the mass of the oxygen vessel. So far, the total mass
of the entire fuel cell UAV has been obtained. Finally, we need
to use the total mass obtained above to verify whether the rated
power of the fuel cell meets the requirements and whether the
Fig. 16. Design flowchart of low-altitude fuel cell UAV [96]. total mass of the UAV is below the upper limit. If so, the whole
process will end. If not, the iterative calculation is required.
Based on the above steps, a detailed design procedure for a
system and the total mass of the UAV should be verified. Fifth, light fuel cell UAV that can achieve an altitude of 10 km can
appropriate flight strategies should be carefully considered be found in [19].
based on the predesigned flight endurance and other related
specifications. Through this step, the amount and mass of
hydrogen fuel can be determined. In terms of oxygen supply, B. Specific Design for Low-Altitude Fuel Cell UAV
there are two commonly used methods: air compressor and Low-altitude UAVs prefer better dynamic performance
pure oxygen vessel. Comparatively speaking, the latter is more and even do some stunts, such as rolls, spins, loops, and
4134 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

Fig. 19. Synthetic roadmap of the integrated functional design framework for the propulsion system identification [98].

Immelmann turns. Therefore, it is important to select an appro- the cruise stage can be obtained separately. Next, we need
priate fuel cell system and battery pack. The function of the to carefully select or design the powertrain structure, fuel
battery is not only the power source of the controller board but cell stack system, battery pack, and even hydrogen storage
also the power source of the propulsion system. The design methods. Fifth, the mass of the entire fuselage can be calcu-
procedure of the low-altitude fuel cell UAVs is given by a lated to verify the feasibility of the power plant. If so, the
flowchart, as shown in Fig. 16. payload can be calculated. If not, the iterative calculation is
First, specific aircraft technical data should be given, such as required. To better meet the dynamic operation, after the above
wingspan, wing load, size, and maximum take-off load. Then, conceptual design, some ground tests are needed, including
a suitable airframe needs to be considered, including wings, stack and hydrogen production tests, and battery charging and
structure, and materials. It may have a great influence on the discharging tests. Based on the above design steps, the detailed
final flight performance. Third, according to the principles of design procedure for low-altitude fuel cell UAVs can be
aerodynamics, the power required for the climb stage and found in [96].
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4135

C. Specific Design for Long-Endurance Fuel Cell UAV constraints related to the technical requirement, normative
For long-endurance fuel cell UAVs, sufficient energy is an requirement, and custom requirement should also be consid-
important factor. Therefore, it is normal that a conceptual ered. Through analyzing the design process of different UAVs,
design process is executed for energy. Fig. 17 gives a brief it can be concluded that there are three main design stages:
flowchart for designing a long-endurance fuel cell UAV. It can 1) the identification of the requirements for the overall UAV;
be seen that it can be divided into three stages: missions and 2) the selection of the actuation systems; and 3) the sizing of
other parameters, constraint and mission analysis, feasibility the power supply subsystem. A detailed explanation of these
verification, and design specifications. To illustrate this process three stages is given in Fig. 19 [98]. From this figure, it can be
more clearly, a detailed flowchart is given, as shown in Fig. 18. found that every stage has its main tasks. The first stage is to
First, similar to other specific fuel cell UAVs, the main char- determine the actuation and set of possible actuator candidates
acteristics of the UAV should be given, including total mass, through the demand analysis of the overall UAV. The second
flight altitude, and flight duration. Then, the input parameters stage is to determine the best candidate for propulsion config-
of the constraint analysis can be determined empirically. Gen- uration and make the first attempt at the energy consumption
erally, these parameters include aspect ratio, efficiency, fric- profile. The final stage is to determine the final functional
tion coefficient, speed ratio, and so on. Third, the parameters parameter set for the propulsion system. Although there are
obtained in the second stage are used for constraint analysis, many different applications and categories of UAVs, they all
which will determine the wing load and power-to-weight ratio need to comply with the abovementioned design stages. The
of the UAV. In practice, wing load and power-to-weight ratio difference is that each type of UAV has a different design
are the two most important factors for UAVs, and they have a focus.
great influence on flight endurance and dynamic performance.
Therefore, the aerodynamic design of the airframe should be V. C ONCLUSION
better, and the choice of fuel cell system should be more cau-
tious. Next, we can investigate the effect of gravimetric power Fuel cell UAVs can be regarded as one of the most
and energy densities on endurance and calculate these two emerging technologies in the world. They may cause tremen-
values. Finally, mission analysis can be performed based on dous changes in the economy and society and greatly
the parameters obtained above. If all the requirements are met, improve people’s lives. Compared with other types of UAVs,
the design specifications for fuel cell UAVs can be determined. fuel cell-powered UAVs have many advantages, which have
If not, iterative computations will be needed. Based on the attracted the attention of many researchers. However, due to
above design steps, the detailed design procedure of the long- the inherent disadvantages of fuel cell systems and some
endurance fuel cell UAV can be found in [97]. technological bottlenecks of the fuel cell electric propulsion
First, similar to other specific fuel cell UAVs, the main char- systems, it is still difficult for fuel cell UAVs to have some
acteristics of the UAV should be given, including total mass, satisfactory characteristics, including long endurance, high
flight altitude, and flight duration. Then, the input parameters reliability, and extended lifespan.
of the constraint analysis can be determined empirically. Gen- In this work, we give a comprehensive review of some
erally, these parameters include aspect ratio, efficiency, fric- key technologies for the electric propulsion system of fuel
tion coefficient, speed ratio, and so on. Third, the parameters cell UAVs. As one of the most important technologies, the
obtained in the second stage are used for constraint analysis, electric propulsion architecture of the fuel cell UAVs is first
which will determine the wing load and power-to-weight ratio given. The pure fuel cell architecture is simple to implement.
of the UAV. In practice, wing load and power-to-weight ratio Nevertheless, the overall efficiency and dynamic response
are the two most important factors for UAVs, and they have a of the pure fuel cell system are limited. Therefore, hybrid
great influence on flight endurance and dynamic performance. propulsion architectures are reviewed, including series hybrid
Therefore, the aerodynamic design of the airframe should be architecture and parallel hybrid architecture. These hybrid
better, and the choice of fuel cell system should be more architectures can combine the advantages of different power
cautious. Next, we can investigate the effect of gravimetric sources. However, due to the characteristics of the two-stage
power and energy densities on endurance and calculate these conversion, the energy conversion efficiency of the series
two values. Finally, mission analysis can be performed based hybrid system is low. By contrast, the parallel hybrid system
on the parameters obtained above. If all requirements are met, is more efficient, but effective EMS is needed. As can be seen,
the design specifications for fuel cell UAVs can be determined. the rule-based EMS methods are mostly used for fuel cell UAV
If not, iterative computations will be needed. Based on the applications. Although many advanced EMS methods have
above design steps, the detailed design procedure of the long- been proposed aiming to improve the flight endurance and
endurance fuel cell UAV can be found in [97]. reliability of fuel cell UAVs, online EMS is usually difficult
to be guaranteed because of the large computational burden.
In addition, most of the proposed advanced methods are just
D. Discussion and Future Prospect verified by simulation or HIL. Therefore, more ground or
In general, the design of the propulsion system of fuel real flight tests with advanced EMS should be investigated
cell UAVs should integrate different kinds of requirements. to prompt the real implementation of EMS on UAVs. As for
These requirements often include the demand of mission pro- the different kinds of fuel cell UAVs, some detailed design
file, power supply, and actuation subsystems. Besides, the procedures are given. In order to provide a design guideline
4136 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

for most kinds of fuel cell UAVs, a general systematic method [7] Global Military UAV Market Size to Hit $13bn by 2024 and
is also provided. Each kind of UAV could be successfully Global Consumer UAV Market Size to Exceed $9bn by 2024.
Accessed: Sep. 30, 2019. [Online]. Available: https://www.gminsights.
designed by focusing on different aspects. According to the com/pressrelease/unmanned-aerial-vehicles-UAV-commercial-drone-
abovementioned analysis, the main conclusions can be made market
as follows. [8] E. Bongermino, F. Mastrorocco, M. Tomaselli, V. G. Monopoli, and
1) Among kinds of propulsion architectures, the parallel D. Naso, “Model and energy management system for a parallel hybrid
electric unmanned aerial vehicle,” in Proc. IEEE 26th Int. Symp. Ind.
hybrid power system shows a greater advantage. How- Electron. (ISIE), Jun. 2017, pp. 1868–1873.
ever, it is dependent on specific situations to select active [9] T. Donateo and L. Spedicato, “Fuel economy of hybrid electric flight,”
or passive parallel hybrid power systems. Appl. Energy, vol. 206, no. 15, pp. 723–738, Nov. 2017.
2) Optimization and intelligent-based EMS methods show [10] I. Burch and J. Gilchrist, “Survey of global activity to phase out internal
combustion engine vehicles,” Center Climate Protection, Santa Rosa,
more attractive prospects. However, the corresponding CA, USA, Tech. Rep., Sep. 2018.
solving algorithms and hardware computation platform [11] M. A. Hannan, F. A. Azidin, and A. Mohamed, “Hybrid electric
need to be developed. vehicles and their challenges: A review,” Renew. Sustain. Energy Rev.,
vol. 29, pp. 135–150, Jan. 2014.
3) For small and medium fuel cell UAVs, there is a general [12] A. Gong and D. Verstraete, “Fuel cell propulsion in small fixed-wing
design procedure to follow. However, the UAVs with unmanned aerial vehicles: Current status and research needs,” Int.
specific requirements need to emphasize some processes J. Hydrogen Energy, vol. 42, no. 33, pp. 21311–21333, Aug. 2017.
while ignoring others. [13] D. Verstraete, K. Lehmkuehler, and K. C. Wong, “Design of a fuel
cell powered blended wing body UAV,” in Proc. ASME Int. Mech.
Furthermore, in the race for widespread applications of fuel Eng. Congr. Expo., 2012, pp. 621–629.
cell UAVs and considering each technical challenge involved [14] APOLLO 13 the Seventh Mission: The Third Lunar Landing
in the development of fuel cell electric propulsion systems, Attempt. Accessed: Apr. 1970. [Online]. Available: https://www.
several critical issues and future directions are also listed as nasa.gov/topics/technology/hydrogen/hydrogen_2009.html
[15] Ó. González-Espasandín, T. J. Leo, and E. Navarro-Arévalo, “Fuel
follows: cells: A real option for unmanned aerial vehicles propulsion,” Sci.
1) to propose innovative hybrid propulsion systems for fuel World J., vol. 2014, pp. 1–12, Jan. 2014.
cell UAVs, considering the mini, medium, and large [16] O. Z. Sharaf and M. F. Orhan, “An overview of fuel cell technology:
UAVs, such that the propulsion efficiency and reliability Fundamentals and applications,” Renew. Sustain. Energy Rev., vol. 32,
pp. 810–853, Apr. 2014.
can be improved; [17] J. Renau et al., “Use of fuel cell stacks to achieve high altitudes in light
2) to develop a high power density and high-reliability unmanned aerial vehicles,” Int. J. Hydrogen Energy, vol. 40, no. 42,
converter for fuel cell UAVs, such that the total take-off pp. 14573–14583, Nov. 2015.
weight of fuel cell UAVs can be reduced; [18] J. Wang, “Barriers of scaling-up fuel cells: Cost, durability and relia-
bility,” Energy, vol. 80, pp. 509–521, Feb. 2015.
3) to develop advanced online EMSs for fuel cell UAVs, [19] W. R. W. Daud, R. E. Rosli, E. H. Majlan, S. A. A. Hamid,
considering different applications, such that global opti- R. Mohamed, and T. Husaini, “PEM fuel cell system control: A review,”
mal results can be obtained; Renew. Energy, vol. 113, pp. 620–638, Dec. 2017.
4) to design more efficient algorithms or develop a more [20] Y. Bicer and I. Dincer, “Life cycle evaluation of hydrogen and other
potential fuels for aircrafts,” Int. J. Hydrogen Energy, vol. 42, no. 16,
powerful computation platform, such that advanced pp. 10722–10738, Apr. 2017.
EMS methods can be implemented in real time; [21] E. Baharozu, G. Soykan, and M. B. Ozerdem, “Future aircraft con-
5) to develop more detailed design guidelines for all kinds cept in terms of energy efficiency and environmental factors,” Energy,
vol. 140, pp. 1368–1377, Dec. 2017.
of fuel cell UAVs, such that as much as possible con- [22] M. D. Fernandes et al., “SOFC-APU systems for aircraft: A review,”
straints can be considered in the design process; Int. J. Hydrogen Energy, vol. 43, no. 33, pp. 16311–16333, Aug. 2018.
6) to research advanced simulation and computation plat- [23] M. N. Boukoberine, Z. Zhou, and M. Benbouzid, “A critical review
forms for the propulsion system of fuel cell UAVs, on unmanned aerial vehicles power supply and energy management:
Solutions, strategies, and prospects,” Appl. Energy, vol. 255, Dec. 2019,
such that the whole design process can be accomplished Art. no. 113823.
automatically. [24] T. H. Bradley, B. A. Moffitt, D. N. Mavris, and D. E. Parekh, “Develop-
ment and experimental characterization of a fuel cell powered aircraft,”
R EFERENCES J. Power Sources, vol. 171, no. 2, pp. 793–801, 2007.
[25] K. Swider-Lyons, M. Schuette, R. Stroman, J. Rodgers, G. Page, and
[1] P. Liu et al., “A review of rotorcraft unmanned aerial vehicle (UAV) J. Mackrell, “Liquid hydrogen fuel system for small unmanned air
developments and applications in civil engineering,” Smart Struct. Syst, vehicles,” in Proc. 51st AIAA Aerosp. Sci. Meeting including Horizons
vol. 13, no. 6, pp. 1065–1094, 2014. Forum Aerosp. Expo., Jan. 2013, p. 467.
[2] D. W. Casbeer, S.-M. Li, R. W. Beard, R. K. Mehra, and T. W. McLain,
[26] T. Kim, “NaBH4 (sodium borohydride) hydrogen generator with a
“Forest fire monitoring with multiple small UAVs,” in Proc. Amer.
volume-exchange fuel tank for small unmanned aerial vehicles powered
Control Conf., 2005, pp. 3530–3535.
by a PEM (proton exchange membrane) fuel cell,” Energy, vol. 69,
[3] W. Jin, H. Ge, H. Du, and X. Xu, “A review on unmanned aerial
pp. 721–727, May 2014.
vehicle remote sensing and its application,” Remote Sens. Inf., vol. 1,
pp. 88–92, 2009. [27] H. Tawfik, Y. Hung, and D. Mahajan, “Metal bipolar plates for PEM
[4] M. Hassanalian and A. Abdelkefi, “Classifications, applications, and fuel cell-a review,” J. Power Sources, vol. 163, no. 2, pp. 755–767,
design challenges of drones: A review,” Progr. Aerosp. Sci., vol. 91, 2007.
pp. 99–131, May 2017. [28] A. Emadi, K. Rajashekara, S. S. Williamson, and S. M. Lukic, “Topo-
[5] H. Shakhatreh et al., “Unmanned aerial vehicles (UAVs): A survey on logical overview of hybrid electric and fuel cell vehicular power system
civil applications and key research challenges,” IEEE Access, vol. 7, architectures and configurations,” IEEE Trans. Veh. Technol., vol. 54,
pp. 48572–48634, 2019. no. 3, pp. 763–770, May 2005.
[6] M. N. Boukoberine, Z. Zhou, and M. Benbouzid, “A critical review [29] R. S. Wegeng, L. R. Pederson, W. E. TeGrotenhuis, and G. A. Whyatt,
on unmanned aerial vehicles power supply and energy management: “Compact fuel processors for fuel cell powered automobiles based on
Solutions, strategies, and prospects,” Appl. Energy, vol. 255, Dec. 2019, microchannel technology,” Fuel Cells Bull., vol. 3, no. 28, pp. 8–13,
Art. no. 113823. Jan. 2001.
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4137

[30] F. G. Harmon, A. A. Frank, and J.-J. Chattot, “Conceptual design and [51] Z. Yang, T. Lei, Z. Lin, H. Fu, and X. Zhang, “The testing platform of
simulation of a small hybrid-electric unmanned aerial vehicle,” J. Aircr., hybrid electric power system for a fuel cell unmanned aerial vehicle,”
vol. 43, no. 5, pp. 1490–1498, Sep. 2006. in Proc. IEEE Int. Conf. Electr. Syst. Aircr., Railway, Ship Propuls.
[31] C. Kim, E. NamGoong, S. Lee, T. Kim, and H. Kim, “Fuel economy Road Vehicles Int. Transp. Electrific. Conf. (ESARS-ITEC), Nov. 2018,
optimization for parallel hybrid vehicles with CVT,” SAE transactions, pp. 1–8.
vol. no., pp. 2161–2167, 1999. [52] T. Lei, Z. Min, H. Fu, W. Li, and X. Zhang, “Dynamic balanced energy
[32] P. Bubna, D. Brunner, J. J. Gangloff, S. G. Advani, and A. K. Prasad, management strategies for fuel cell hybrid power of unmanned air
“Analysis, operation and maintenance of a fuel cell/battery series- vehicle,” (in Chinese), Acta Aeronautica et Astronautica Sinica, vol. 41,
hybrid bus for urban transit applications,” J. Power Sources, vol. 195, no. 12, Dec. 2020, Art. no. 324048.
no. 12, pp. 3939–3949, Jun. 2010. [53] L. Karunarathne, J. T. Economou, and K. Knowles, “Power and energy
[33] F. Re, “Viability and state of the art of environmentally friendly aircraft management system for fuel cell unmanned aerial vehicle,” Proc.
taxiing systems,” in Proc. Electr. Syst. Aircr., Railway Ship Propuls., Inst. Mech. Eng., G, J. Aerosp. Eng., vol. 226, no. 4, pp. 437–454,
2012, pp. 1–6. Apr. 2012.
[34] M. Harmats and D. Weihs, “Hybrid-propulsion high-altitude long- [54] X. Zhang, L. Liu, Y. Dai, and T. Lu, “Experimental investigation
endurance remotely piloted vehicle,” J. Aircr., vol. 36, no. 2, on the online fuzzy energy management of hybrid fuel cell/battery
pp. 321–331, Mar. 1999. power system for UAVs,” Int. J. Hydrogen Energy, vol. 43, no. 21,
[35] Z. Ji, J. Qin, K. Cheng, F. Guo, S. Zhang, and P. Dong, “Thermo- pp. 10094–10103, May 2018.
dynamics analysis of a turbojet engine integrated with a fuel cell and [55] B. Omar, A. Savvaris, R. O. Abdulhadi, M. K. Afdhol, and
steam injection for high-speed flight,” Energy, vol. 185, pp. 190–201, M. Hasibuan, “Saving hydrogen fuel consumption and operating at
Oct. 2019. high efficiency of fuel cell in hybrid system to power UAV,” J. Earth
[36] P. Aguiar, D. Brett, and N. Brandon, “Solid oxide fuel cell/gas turbine Energy Eng., vol. 10, no. 1, pp. 32–42, Mar. 2021.
hybrid system analysis for high-altitude long-endurance unmanned aer- [56] Y. Shen, P. Cui, X. Wang, X. Han, and Y.-X. Wang, “Variable structure
ial vehicles,” Int. J. Hydrogen Energy, vol. 33, no. 23, pp. 7214–7223, battery-based fuel cell hybrid power system and its incremental fuzzy
Dec. 2008. logic energy management strategy,” Int. J. Hydrogen Energy, vol. 45,
no. 21, pp. 12130–12142, Apr. 2020.
[37] Z. Ji, J. Qin, K. Cheng, C. Dang, S. Zhang, and P. Dong, “Thermo-
dynamic performance evaluation of a turbine-less jet engine integrated [57] A. Ravey, B. Blunier, and A. Miraoui, “Control strategies for fuel-cell-
with solid oxide fuel cells for unmanned aerial vehicles,” Appl. Thermal based hybrid electric vehicles: From offline to online and experimental
Eng., vol. 160, Sep. 2019, Art. no. 114093. results,” IEEE Trans. Veh. Technol., vol. 61, no. 6, pp. 2452–2457,
Jul. 2012.
[38] K. Rajashekara, J. Grieve, and D. Daggett, “Hybrid fuel cell power
[58] X. Zhang, L. Liu, and Y. Dai, “Fuzzy state machine energy manage-
in aircraft,” IEEE Ind. Appl. Mag., vol. 14, no. 4, pp. 54–60,
ment strategy for hybrid electric UAVs with PV/fuel cell/battery power
Jul. 2008.
system,” Int. J. Aerosp. Eng., vol. 2018, pp. 1–16, Jul. 2018.
[39] E. L. González et al., “Experimental evaluation of a passive fuel
[59] M. Gadalla and S. Zafar, “Analysis of a hydrogen fuel cell-PV power
cell/battery hybrid power system for an unmanned ground vehi-
system for small UAV,” Int. J. Hydrogen Energy, vol. 41, no. 15,
cle,” Int. J. Hydrogen Energy, vol. 44, no. 25, pp. 12772–12782,
pp. 6422–6432, Apr. 2016.
May 2019.
[60] E. Ozbek, G. Yalin, M. U. Karaoglan, S. Ekici, C. O. Colpan, and
[40] J. Bernard et al., “Fuel cell/battery passive hybrid power source
T. H. Karakoc, “Architecture design and performance analysis of a
for electric powertrains,” J. Power Sources, vol. 196, no. 14, hybrid hydrogen fuel cell system for unmanned aerial vehicle,” Int.
pp. 5867–5872, Jul. 2011. J. Hydrogen Energy, vol. 46, no. 30, pp. 16453–16464, Apr. 2021.
[41] K. Aoki, S. Kuroda, S. Kajiwara, H. Sato, and Y. Yamamoto, “Devel- [61] B. Vural et al., “Fuel cell and ultra-capacitor hybridization: A prototype
opment of integrated motor assist hybrid system: Development of test bench based analysis of different energy management strategies
the ‘insight’ a personal hybrid coupe,” SAE Trans., vol. 109, no. 3, for vehicular applications,” Int. J. Hydrogen Energy, vol. 35, no. 20,
pp. 1883–1890, Jan. 2000. pp. 11161–11171, Oct. 2010.
[42] D. Verstraete, A. Gong, D. D.-C. Lu, and J. L. Palmer, “Experimental [62] S. Bayram and A. R. Boynuegri, “A comperative study on Wiener
investigation of the role of the battery in the AeroStack hybrid, fuel- filter and wavelet transform for energy management systems on hybrid
cell-based propulsion system for small unmanned aircraft systems,” Int. unmanned aerial vehicles,” in Proc. 6th Int. Conf. Electr. Power Energy
J. Hydrogen Energy, vol. 40, no. 3, pp. 1598–1606, Jan. 2015. Convers. Syst. (EPECS), Oct. 2020, pp. 112–117.
[43] A. Nishizawa, J. Kallo, O. Garrot, and J. Weiss-Ungethüm, “Fuel cell [63] M. N. Boukoberine, M. F. Zia, M. Benbouzid, Z. Zhou, and
and li-ion battery direct hybridization system for aircraft applications,” T. Donateo, “Hybrid fuel cell powered drones energy management
J. Power Sources, vol. 222, pp. 294–300, Jan. 2013. strategy improvement and hydrogen saving using real flight test data,”
[44] J. Renau, F. Sánchez, A. Lozano, J. Barroso, and F. Barreras, “Analysis Energy Convers. Manage., vol. 236, May 2021, Art. no. 113987.
of the performance of a passive hybrid powerplant to power a light- [64] S. N. Motapon, L.-A. Dessaint, and K. Al-Haddad, “A comparative
weight unmanned aerial vehicle for a high altitude mission,” J. Power study of energy management schemes for a fuel-cell hybrid emergency
Sources, vol. 356, pp. 124–132, Jul. 2017. power system of more-electric aircraft,” IEEE Trans. Ind. Electron.,
[45] W. Jiang and B. Fahimi, “Active current sharing and source man- vol. 61, no. 3, pp. 1320–1334, Mar. 2014.
agement in fuel cell–battery hybrid power system,” IEEE Trans. Ind. [65] T. Bradley, B. Moffitt, D. Parekh, T. Fuller, and D. Mavris, “Energy
Electron., vol. 57, no. 2, pp. 752–761, Feb. 2009. management for fuel cell powered hybrid-electric aircraft,” in Proc. 7th
[46] B. Wang et al., “Current technologies and challenges of applying fuel Int. Energy Convers. Eng. Conf., Aug. 2009, p. 4590.
cell hybrid propulsion systems in unmanned aerial vehicles,” Prog. [66] X. Li, Y. Wang, D. Yang, and Z. Chen, “Adaptive energy management
Aerosp. Sci., vol. 116, Jul. 2020, Art. no. 100620. strategy for fuel cell/battery hybrid vehicles using Pontryagin’s minimal
[47] B. G. Gang and S. Kwon, “Design of an energy management technique principle,” J. Power Sources, vol. 440, Nov. 2019, Art. no. 227105.
for high endurance unmanned aerial vehicles powered by fuel and solar [67] X. Sun, Y. Zhou, L. Huang, and J. Lian, “A real-time PMP energy
cell systems,” Int. J. Hydrogen Energy, vol. 43, no. 20, pp. 9787–9796, management strategy for fuel cell hybrid buses based on driving seg-
May 2018. ment feature recognition,” Int. J. Hydrogen Energy, vol. 46, no. 80,
[48] H. Rezk, A. M. Nassef, M. A. Abdelkareem, A. H. Alami, and A. Fathy, pp. 39983–40000, 2021.
“Comparison among various energy management strategies for reduc- [68] X. Meng, Q. Li, W. Chen, and G. Zhang, “An energy management
ing hydrogen consumption in a hybrid fuel cell/supercapacitor/battery method based on pontryagin minimum principle satisfactory opti-
system,” Int. J. Hydrogen Energy, vol. 46, no. 8, pp. 6110–6126, mization for fuel cell hybrid systems,” Proc. CSEE, vol. 39, no. 3,
Jan. 2021. pp. 782–792, Feb. 2019.
[49] C. De Wagter et al., “The NederDrone: A hybrid lift, hybrid [69] Z. Cheng, H. Liu, P. Yu, L. Zhu, and T. Sun, “Energy management for
energy hydrogen UAV,” Int. J. Hydrogen Energy, vol. 46, no. 29, fuel cell/battery hybrid unmanned aerial vehicle,” Int. J. Electrochem.
pp. 16003–16018, Apr. 2021. Sci., vol. 16, no. 10, pp. 1–17, Oct. 2021.
[50] B. Lee, S. Kwon, P. Park, and K. Kim, “Active power management [70] Y. Xie, A. Savvaris, and A. Tsourdos, “Fuzzy logic based equivalent
system for an unmanned aerial vehicle powered by solar cells, a fuel consumption optimization of a hybrid electric propulsion system for
cell, and batteries,” IEEE Trans. Aerosp. Electron. Syst., vol. 50, no. 4, unmanned aerial vehicles,” Aerosp. Sci. Technol., vol. 85, pp. 13–23,
pp. 3167–3177, Oct. 2014. Feb. 2019.
4138 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 8, NO. 4, DECEMBER 2022

[71] H. Li, A. Ravey, A. N’Diaye, and A. Djerdir, “A novel equivalent [94] A. Himansu, J. E. Freeh, C. J. Steffen, R. T. Tornabene, and
consumption minimization strategy for hybrid electric vehicle powered X.-Y. J. Wang, “Hybrid solid oxide fuel cell/gas turbine sys-
by fuel cell, battery and supercapacitor,” J. Power Sources, vol. 395, tem design for high altitude long endurance aerospace missions,”
pp. 262–270, Aug. 2018. Tech. Rep. NASA/TM-2006-214328, 2006.
[72] Y. Zhou, A. Ravey, and M.-C. Péra, “Multi-mode predictive energy [95] B. A. Moffitt, T. H. Bradley, D. E. Parekh, and D. N. Mavris, “Design
management for fuel cell hybrid electric vehicles using Markov driving space exploration of small-scale PEM fuel cell unmanned aerial vehi-
pattern recognizer,” Appl. Energy, vol. 258, Jan. 2020, Art. no. 114057. cle,” in Proc. AIAA, Sep. 2006, Paper 7701.
[73] C. Sun, X. Hu, S. J. Moura, and F. Sun, “Velocity predictors for [96] N. Lapeña-Rey, J. A. Blanco, E. Ferreyra, J. L. Lemus, S. Pereira,
predictive energy management in hybrid electric vehicles,” IEEE Trans. and E. Serrot, “A fuel cell powered unmanned aerial vehicle for low
Control Syst. Technol., vol. 23, no. 3, pp. 1197–1204, May 2015. altitude surveillance missions,” Int. J. Hydrogen Energy, vol. 42, no. 10,
[74] R. Zhang, J. Tao, and H. Zhou, “Fuzzy optimal energy management pp. 6926–6940, Mar. 2017.
for fuel cell and supercapacitor systems using neural network based [97] T. H. Oh, “Conceptual design of small unmanned aerial vehicle with
driving pattern recognition,” IEEE Trans. Fuzzy Syst., vol. 27, no. 1, proton exchange membrane fuel cell system for long endurance mis-
pp. 45–57, Jan. 2019. sion,” Energy Convers. Manage., vol. 176, pp. 349–356, Nov. 2018.
[75] D. Min, Z. Song, H. Chen, T. Wang, and T. Zhang, “Genetic algo- [98] B. Wang et al., “Current technologies and challenges of applying fuel
rithm optimized neural network based fuel cell hybrid electric vehicle cell hybrid propulsion systems in unmanned aerial vehicles,” Prog.
energy management strategy under start-stop condition,” Appl. Energy, Aerosp. Sci., vol. 116, Jul. 2020, Art. no. 100620.
vol. 306, Jan. 2022, Art. no. 118036. [99] F. G. Harmon, “Neural network control of a parallel hybrid-electric
[76] J. Yuan, L. Yang, and Q. Chen, “Intelligent energy management strat- propulsion system for a small unmanned aerial vehicle,” Ph.D. disser-
egy based on hierarchical approximate global optimization for plug-in tation, Dept. Mech. Aeronaut. Eng., Univ. California-Davis, Davis, CA,
fuel cell hybrid electric vehicles,” Int. J. Hydrogen Energy, vol. 43, USA, 2005.
no. 16, pp. 8063–8078, Apr. 2018. [100] Q. Li, X. Meng, F. Gao, G. Zhang, and W. Chen, “Approximate
[77] H. Sun, Z. Fu, F. Tao, L. Zhu, and P. Si, “Data-driven reinforcement- cost-optimal energy management of hydrogen electric multiple unit
learning-based hierarchical energy management strategy for fuel trains using double Q-learning algorithm,” IEEE Trans. Ind. Electron.,
cell/battery/ultracapacitor hybrid electric vehicles,” J. Power Sources, vol. 69, no. 9, pp. 9099–9110, Sep. 2022.
vol. 455, Apr. 2020, Art. no. 227964. [101] Z. Yu, D. Zinger, and A. Bose, “An innovative optimal power alloca-
[78] Y. Zhang, R. Ma, D. Zhao, Y. Huangfu, and W. Liu, “A novel energy tion strategy for fuel cell, battery and supercapacitor hybrid electric
management strategy based on dual reward function Q-learning for fuel vehicle,” J. Power Sources, vol. 196, no. 4, pp. 2351–2359, 2011.
cell hybrid electric vehicle,” IEEE Trans. Ind. Electron., vol. 69, no. 2, [102] J. Zhao, X. Li, C. Shum, and J. McPhee, “A review of physics-based
pp. 1537–1547, Feb. 2022. and data-driven models for real-time control of polymer electrolyte
[79] H. Shen, Y. Zhang, J. Mao, Z. Yan, and L. Wu, “Energy management membrane fuel cells,” Energy AI, vol. 6, Dec. 2021, Art. no. 100114.
of hybrid UAV based on reinforcement learning,” Electronics, vol. 10, [103] D. Oryshchyn, N. F. Harun, D. Tucker, K. M. Bryden, and L. Shadle,
no. 16, p. 1929, Aug. 2021. “Fuel utilization effects on system efficiency in solid oxide fuel cell
[80] M. A. Latif, “An agricultural perspective on flying sensors: State of gas turbine hybrid systems,” Appl. Energy, vol. 228, pp. 1953–1965,
the art, challenges, and future directions,” IEEE Geosci. Remote Sens. Oct. 2018.
Mag., vol. 6, no. 4, pp. 10–22, Jul. 2018. [104] Q. Li, X. Meng, F. Gao, G. Zhang, W. Chen, and K. Rajashekara,
[81] G. Potrino, N. Palmieri, V. Antonello, and A. Serianni, “Drones support “Reinforcement learning energy management for fuel cell hybrid sys-
in precision agriculture for fighting against parasites,” in Proc. 26th tem: A review,” IEEE Ind. Electron. Mag., early access, Feb. 21, 2022,
Telecommun. Forum (TELFOR), Nov. 2018, pp. 1–4. doi: 10.1109/MIE.2022.3148568.
[82] R. Ke, Z. Li, S. Kim, J. Ash, Z. Cui, and Y. Wang, “Real-time bidi- [105] Z. Hua, Z. Zheng, E. Pahon, M.-C. Péra, and F. Gao, “A review on
rectional traffic flow parameter estimation from aerial videos,” IEEE lifetime prediction of proton exchange membrane fuel cells system,”
Trans. Intell. Transp. Syst., vol. 18, no. 4, pp. 890–901, Apr. 2016. J. Power Sources, vol. 529, May 2022, Art. no. 231256.
[83] Commercial UAV Expo Americas Virtual Event, To Be Held [106] Z. Gong et al., “A 1 + 1-D multiphase proton exchange membrane fuel
September 15–17. Accessed: Aug. 19, 2020. [Online]. Available: cell model for real-time simulation,” IEEE Trans. Transport. Electrific.,
https://insidetowers.com/cell-tower-news-att-working-make-drone- vol. 8, no. 2, pp. 2928–2944, Jun. 2022.
tower-inspections-intelligent/ [107] Z. Hua, Z. Zheng, M.-C. Péra, and F. Gao, “Statistical analysis on
[84] F. Mohamadi, “Vertical takeoff and landing (VTOL) small random matrices of echo state network in PEMFC system’s lifetime
unmanned aerial system for monitoring oil and gas pipelines,” prediction,” Appl. Sci., vol. 12, no. 7, p. 3421, Mar. 2022.
U.S. Patent 8 880 241, Nov. 4, 2014.
[85] K. P. Valavanis and G. J. Vachtsevanos, Handbook of Unmanned Aerial
Vehicles. Berlin, Germany: Springer, 2015. Liangcai Xu (Member, IEEE) received the B.Eng.
[86] F. P. G. Márquez and I. S. Ramírez, “Condition monitoring system for degree in measurement and control technology and
solar power plants with radiometric and thermographic sensors embed- instrument and the master’s degree in electrical engi-
ded in unmanned aerial vehicles,” Measurement, vol. 139, pp. 152–162, neering from Northwestern Polytechnical University,
Jun. 2019. Xi’an, China, in 2018 and 2021, respectively. He is
[87] L. Wang and Z. Zhang, “Automatic detection of wind turbine blade currently pursuing the Ph.D. degree with the Depart-
surface cracks based on UAV-taken images,” IEEE Ind. Electron. Mag., ment of Mechanical Engineering, National Univer-
vol. 64, no. 9, pp. 7293–7303, Sep. 2017. sity of Singapore, Singapore.
[88] Z. Zhou, C. Zhang, C. Xu, F. Xiong, Y. Zhang, and T. Umer, “Energy- After graduation, he worked as a Research
efficient industrial internet of UAVs for power line inspection in smart Assistant at Northwestern Polytechnical University
grid,” IEEE Trans. Ind. Informat., vol. 14, no. 6, pp. 2705–2714, from 2021 to 2022. His research interests include
Jun. 2018. advanced and fault-tolerant control for power electronics, and optimization of
[89] H. Abaunza and P. C. Garcia, “Les applications de drones aériens- storage system and its application in power systems.
L’utilisation civile desUAVs,” (in French), Techn. de l’Ingenieur,
Aug. 2017.
[90] T. F. Villa, F. Gonzalez, B. Miljievic, Z. D. Ristovski, and L. Morawska, Yigeng Huangfu (Senior Member, IEEE) received
“An overview of small unmanned aerial vehicles for air quality the M.S. and Ph.D. degrees in electrical engineering
measurements: Present applications and future prospectives,” Sensors, from Northwestern Polytechnical University (NPU),
vol. 16, no. 7, p. 1072, 2016. Xi’an, China, in 2007 and 2009, respectively, and
[91] W. Zang, J. Lin, Y. Wang, and H. Tao, “Investigating small-scale the Ph.D. degree from the University of Technology
water pollution with UAV remote sensing technology,” in Proc. World of Belfort-Montbéliard, Belfort, France, in 2010.
Automat. Congr., 2012, pp. 1–4. He has been a Full Professor with NPU. His main
[92] T. H. Bradley, B. A. Moffitt, T. F. Fuller, D. N. Mavris, and research interests include power electrical conver-
D. E. Parekh, “Comparison of design methods for fuel-cell-powered sion, transportation electrification, intelligence con-
unmanned aerial vehicles,” J. Aircr., vol. 46, no. 6, pp. 1945–1956, trol of new energy conversion, and renewable energy
Nov. 2009. generation technology.
[93] L. Kohout and P. Schmitz, “Fuel cell propulsion systems for an all- Dr. Huangfu is currently the Chair of the IEEE IES Technical Committee
electric personal air vehicle,” in Proc. AIAA Int. Air Space Symp. Expo., on Transportation Electrification. He is also an Associate Editor of IEEE
Next 100 Years, Jul. 2003, p. 2867. Industrial Electronics and Application and IET Power Electronics.
XU et al.: COMPREHENSIVE REVIEW ON FUEL CELL UAV KEY TECHNOLOGIES 4139

Rui Ma (Senior Member, IEEE) received the mas- Dongdong Zhao (Member, IEEE) received the
ter’s degree in electrical engineering from North- B.Eng. degree in information antagonizing tech-
western Polytechnical University (NPU), Xi’an, nology from Northwestern Polytechnical University,
China, in 2016, and the dual Ph.D. degree in electri- Xi’an, China, in 2008, and the Ph.D. degree in elec-
cal engineering from the University of Technology trical engineering from the University of Technology
of Belfort-Montbéliard (UTBM), Belfort, France, of Belfort-Montbéliard, Belfort, France, in 2014.
and NPU, in 2018 and 2019, respectively. Since 2014, he has been an Associate Profes-
He is currently an Associate Professor with the sor with Northwestern Polytechnical University. His
School of Automation, NPU. His main research research interests include fuel cell system modeling
interests include fuel cell modeling, fuel cell appli- and control, power electronics, and motor drives.
cations, real-time simulation systems, and machine
learning.
Dr. Ma served as a member of the Technical Committee on Transportation
Electrification for the IEEE Industrial Electronics Society and the Special
Session Chair for 45th, 46th, and 47th IEEE IECON. He serves as an Associate Yongheng Yang (Senior Member, IEEE) received
Editor for IEEE IES Industrial Electronics Technology News and a Distinguish the B.Eng. degree in electrical engineering and
Reviewer for IEEE T RANSACTIONS ON I NDUSTRIAL E LECTRONICS . automation from Northwestern Polytechnical Uni-
versity, Xi’an, China, in 2009, the master’s degree
from Southeast University, Nanjing, China, in 2011,
and the Ph.D. degree in energy technology (power
electronics and drives) from Aalborg University,
Aalborg, Denmark, in 2014.
In 2013, he was a Visiting Scholar at Texas A&M
University, College Station, TX, USA. From 2014 to
2020, he was with the Department of Energy Tech-
nology, Aalborg University, where he became a tenured Associate Professor
in 2018. In January 2021, he joined Zhejiang University, Hangzhou, China,
Renyou Xie (Member, IEEE) received the bachelor’s as a ZJU100 Professor. His research focuses on the grid integration of
degree in electrical engineering from the School of photovoltaic systems and control of power converters, in particular, the grid-
Automation, Northwestern Polytechnical University forming technologies.
(NPU), Xi’an, China, in 2018, and the master’s Dr. Yang is currently the Secretary of the IEEE PELS Technical Com-
degree in electrical engineering from NPU in 2021. mittee on Sustainable Energy Systems and a Council Member of the China
He is currently a Research Assistant with Cen- Power Supply Society. He was the Chair of the IEEE Denmark Section
tral South University, Changsha, China. His current from 2019 to 2020. He is an associate editor for several IEEE T RANSAC -
research interests include fuel cells, federated learn- TIONS . He was a recipient of the 2018 IET Renewable Power Generation
ing, and renewable energy systems. Premium Award, the 2021 Richard M. Bass Outstanding Young Power Elec-
tronics Engineer Award from the IEEE Power Electronics Society (PELS), and
the 2022 Isao Takahashi Power Electronics Award. He was an Outstanding
Reviewer of the IEEE T RANSACTIONS ON P OWER E LECTRONICS in 2018.
He received two IEEE best paper awards. He has been included on the list
of Highly Cited Chinese Researchers by Elsevier in 2022.

Yupeng Wang (Senior Member, IEEE) received

the M.E. degree in electrical engineering from the
University of Technology of Belfort-Montbéliard
Ziyou Song (Member, IEEE) received the B.E. (UTBM), Belfort, France, in 2013.
and Ph.D. degrees (Hons.) in automotive engineer- He is currently the Team Leader of the fuel cell
ing from Tsinghua University, Beijing, China, in system performance with FAW, Changchun, China.
2011 and 2016, respectively. He is mainly responsible for the performance design,
After graduation, he worked as a Research Sci- strategy development, virtual simulation, and data
entist at Tsinghua University from 2016 to 2017. analysis of the fuel cell system.
From 2017 to 2019, he worked as a Post-Doctoral
Research Fellow at the University of Michigan,
Ann Arbor, MI, USA, where he was an Assistant
Research Scientist/Lecturer from 2019 to 2020. He is
currently an Assistant Professor with the Department Liangfei Xu received the B.E. and Ph.D. degrees
of Mechanical Engineering, National University of Singapore, Singapore. from Tsinghua University, Beijing, China, in
Prior to joining NUS, he was a Battery Algorithm Engineer at Apple Inc., 2003 and 2009, respectively.
Cupertino, CA, USA, where he was in charge of battery management system From November 2015 to August 2017, he was
for audio products (e.g., AirPods). He is the author or the coauthor of two with Forschungszentrum Jülich, Jülich, Germany,
book chapters and more than 60 peer-reviewed publications, including 40 jour- as an Alexander von Humboldt Research Fellow.
nal articles. His research interests lie in the areas of modeling, estimation, In October 2017, he was promoted to an Associate
optimization, and control of energy storage (e.g., battery, supercapacitor, and Professor at the School of Vehicle and Mobility,
flywheel) for electrified vehicles and renewable energy systems. Using energy Tsinghua University. He is currently an Associate
storage as a bridge, his group is actively connecting the automotive, trans- Professor and the Deputy Director of the School of
portation, and power system communities by collaborating on interdisciplinary Vehicle and Mobility, Vehicular Powertrain Engi-
projects to ensure that both power system and transportation system are neering Institute, Tsinghua University. His research interests include the
efficient, reliable, and clean. design, diagnosis, control, and health management of polymer electrolyte
Dr. Song received several paper awards, including the Applied Energy membrane (PEM) fuel cells and powertrains for electric vehicles.
2015–2016 Highly Cited Paper Award, the Applied Energy Award for Most Dr. Xu serves as an Associate Editor for International Journal of Power-
Cited Energy Article from China, the NSK Outstanding Paper Award of trains and an Editorial Board Member for eTransportation. He received the
Mechanical Engineering, and the 2013 IEEE VPPC Best Student Paper Award. 5th ProSPER.NET-Scopus Young Scientist Award in 2013, the First Prize
He serves as a reviewer for more than 40 international journals. He also serves of China Automotive Industry Science and Technology Award in 2019, the
as an Associate Editor for Automotive Innovation, SAE International Journal Science Fund Program for Distinguished Young Scholars in 2020, the First
of Electrified Vehicles, and IEEE T RANSACTIONS ON T RANSPORTATION Prize of Beijing Science and Technology Award in 2021, and the Outstanding
E LECTRIFICATION. Young Technological Talent Award of China Automobile Industry in 2021.