Clean Energy, 2022, Vol. 6, No.

4, 671–681

6 https://doi.org/10.1093/ce/zkac047
Advance access publication 20 August 2022
Research Article

Unified fuzzy logic controller and power management

for an isolated residential hybrid PV/diesel/battery
energy system
Adil Atoui1,*, , Mohamed Seghir Boucherit1, Khelifa Benmansour2, Said Barkat3, Ali Djerioui3 and Azeddine Houari4
1
Process Control Laboratory, National Polytechnic School, Algiers, Algeria
2
Polytechnic Military school, UER ELT, Algiers, Algeria

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

3
Electrical engineering, University of M’sila, M’sila, Algeria
4
Nantes University, Nantes, France
*Corresponding author. E-mail: [email protected]

Abstract
Hybrid systems based on renewable energies for the electrification of remote sites controlled by power management systems (PMSs)
aim to reduce fossil fuels and increase the efficiency of renewable energy sources to minimize greenhouse gas emissions. The influ-
ential role of the PMS contributes to improving the efficiency and effectiveness of these systems by ensuring a balance between the
different sources and loads in all operating modes. However, the abrupt transitions between the various operational modes selected
by the PMS generate power loss and imbalance. To handle this issue, a fuzzy logic controller (FLC)-based PMS controlling a photovol-
taic (PV) and diesel hybrid system with a battery storage element connected to a DC bus is proposed in this paper. The proposed PMS
is wholly based on FLC to ensure a smooth transition between the different modes of the system. The success of using the suggested
PMS lies in how well the FLC parameters are chosen before the system is processed. For this purpose, the particle swarm optimization
algorithm is adapted to tune the FLC parameters. The resulting optimal intelligent PMS is tested and compared with a classical one
using comprehensive simulations performed in a Simscape ElectricalTM MATLAB® environment. The obtained results show an over-
shoot attenuation at the DC-bus voltage of 2% when changing the mode and an improvement in the PV generator efficiency by 99.5%.

Graphical Abstract
Sensor signal
Command
PV generator Bus DC Power flow
DC

DC

Diesel DC

DC
AC

DC DC

AC

Batteries

DC

DC

µ
NB NM NS Z PS PM PB

FUZZY LOGIC CONTROL

POWER MANGEMENT SYSTEM

Keywords: fuzzy logic control; hybrid power systems; particle swarm optimization; photovoltaic systems; power management system

Received: 5 March 2022. Accepted: 19 July 2022

© The Author(s) 2022. Published by Oxford University Press on behalf of National Institute of Clean-and-Low-Carbon Energy
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/
licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For
commercial re-use, please contact [email protected]
672 | Clean Energy, 2022, Vol. 6, No. 4

Introduction of sunlight and wind speed, which could cause a decrease in the
system performance if these measurements were not accurate
Today, the most remote areas in the world are supplied by electric
enough. In [15], the PMS-FLC was used to control a hybrid system
power systems based on diesel generators due to their availability
whose sources were PV, wind turbine, battery bank and an elec-
and reliability. Unfortunately, their operating efficiency is reduced
trolysis system to produce hydrogen. Without limited operating
due to several factors, such as their operation at low power (<40–
modes, the PV generator always delivered its maximum power
50% of their nominal power) and the high cost of transporting
in this system. The surplus energy would be stored in the bat-
fuel to the remote area [1, 2]. Some environmental concerns, like
teries or used for hydrogen gas production. Other works used
the emission of greenhouse gases (GHGs), give another reason to
the PMS-FLC only to manage hybrid storage units like that pro-
reduce their utilization at a maximum pace [3]. These factors fa-
posed in [18]. The control logic decided the proportion of power
vour hybrid systems as an alternative energy source for isolated
to be shared among the energy-storage devices (ESDs) of a hy-
areas [4, 5]. This involves using several renewable energy sources
brid system with a PV generator and multiple ESDs. In [19, 20],
and energy-storage systems to ensure a certain complementarity,
the PMS-FLC ensured the power requested by the load by provid-
increasing their advantages and minimizing their disadvantages
ing the state of charge of the supercapacitors and the batteries
[6].
within acceptable levels.

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

Consequently, an efficient power management system (PMS)
In this paper, in contrast to the works mentioned above, the
and robust regulation are required to exploit all sources with
PMS and regulators are both based on fuzzy logic with a reduced
maximum efficiency in variable climatic conditions and during
number of rules, which guarantees a smooth transition between
any unpredictable load change [7]. In hybrid systems, connected
all operating modes of the system and enhances its flexibility and
to the grid or not, the different sources and loads are intercon-
efficiency. The PMS-FLC is used to manage three sources (PV, die-
nected through either AC or DC buses or both; this connection is
sel and battery) with the following features:
performed via static converters [8], which will control the energy
flow between their two terminals. These converters are controlled
• the PV system can operate in two modes: MPPT (maximum
by regulators whose references are provided by the PMS [9].
power point tracking), where it delivers its maximum
In the literature, there are many works on the PMS to im-
power, or LPP (limited power point) mode to provide just
prove the quality of the power by ensuring a stable voltage of the
the power required by the load;
bus and achieving a balance between the loads and the sources
• the proposed system does not need a dump load;
operating at their maximum efficiencies, which improves the
• sunlight sensor is no longer required;
overall hybrid system efficiency, regardless of climate change or
• the battery state of charge (SOC) is maintained within the
load variation. In [10], a stand-alone photovoltaic (PV)/diesel/bat-
permissible operating interval;
tery system was managed by a PMS to control the system in two
• the generated reference powers are consistent as required
modes. In isolated mode control, the system used PV and bat-
by the load;
tery to keep the frequency and magnitude of the charge voltage
• the starting and stopping of the diesel generator are soft.
fixed. In parallel mode control, when the battery was discharged,
the diesel operated at its nominal power and with the PV power,
The significant contributions of this work are summar-
giving the power requested by the load. The authors in [11–13]
ized below:
proposed a PMS of a hybrid system (PV and storage) capable of
ensuring the efficiency of the system in all its operating modes.
• To develop an FLC by optimizing its parameters (scaling
They showed that the system could operate in an isolated or grid-
parameters and the membership functions (MFs)) by the
connected mode, leaving the battery state of charge between ac-
particle swarm optimization (PSO) algorithm;
ceptable margins as recommended by the manufacturers. The
• To validate the proposed PMS through MATLAB® simula-
authors in [9] added ultra-capacitors to minimize the battery cur-
tions by comparing it with traditional PMS.
rent peaks during fast changes. In the same direction, another
PMS was proposed in [14] to manage three sources (PV, battery
and diesel) connected to the AC bus and feeding an isolated load. This paper is organized as follows. Section 1 presents a general
In [15], a smart microgrid including PV, diesel and battery stor- description of the hybrid system power structure. The proposed
age was offered with an energy management system to suppress PMS-FLC concept is presented in Section 2. Section 3 discusses
power fluctuations due to PV generators by medium-term storage simulation results by Simscape ElectricalTM of MATLAB® to verify
(battery) or long-term storage (diesel). the proposed control system’s effectiveness. Finally, some conclu-
Although the works mentioned above have different struc- sions are drawn in Section 4.
tures depending on the nature and number of sources involved
in the hybrid system, the PMS is the common point. The flip side
of the PMS is its inherent structure, which is based on a flowchart 1 Description of the system under study
with several sequences and conditions needed to select the ap- Fig. 1 illustrates the configuration of the proposed system, which
propriate mode in which the system will operate. This structure uses a centralized PMS manager. It consists of a PV generator and
imposes sudden changes in the reference values, which presents bank of batteries that are connected to the DC bus via DC/DC
a tremendous challenge for the regulators to keep the desired boost converters, unidirectional for the PV and bidirectional for
power quality and system stability. Recent works have been pro- the batteries; the latter ensures the control of the charge and the
posed in [16, 17] to maintain flexibility and soften abrupt changes discharge of these batteries and the diesel generator, which is
using fuzzy logic-based PMS (PMS-FLC). In [14], the PMS-FLC con- connected with the DC bus via an uncontrolled rectifier (AC/DC)
trolled a hybrid system containing three sources (wind, PV and followed by a DC/DC boost converter. It is a non-grid-connected
diesel) and battery storage feeding an isolated load. The sug- system with a DC bus structure to power an isolated AC load that
gested PMS-FLC was based on the measurement of the intensity simulates a home through a DC/AC inverter.
 Unified fuzzy logic controller and power management for an isolated residential hybrid PV/diesel/battery energy system | 673

PPV

lPV
Vpv Boost
converter
PAC Load

Pbat
lbat Inverter

Vbat
Bidirectional
converter Pac load

SOC
DC

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

Pdies
Load

Rectifier Buck Pdc load

converter

Pdies Vdc BUS

PMS system with controller

Fig. 1: The proposed stand-alone hybrid system.

The PMS is required to ensure effective interaction between • The sources selected to feed the load are planned by order
power sources for an efficient power balance load, protect the of priority; the PV generator is considered as the primary
battery from deep discharging or overcharging and prevent a source, then the battery storage as the secondary and, fi-
blackout. Depending on the situation of the monitored param- nally, the diesel generator.
eters (sunshine, battery charge status and load demand) in real • The battery charging is done only by using the surplus
time, the PMS determines the operating mode of the system and power generated by the PV generator.
reacts to the values given to the various regulator’s converters to • The diesel generator is activated only when the battery is
control the powers of each source. discharged and the power of the PV generator is insufficient.

2 The proposed power management 2.1 PV system control

system using fuzzy logic control (PMS-FLC) There are two operation modes for the PV generator. The first
The review of similar literature demonstrated that the most chal- mode is the MPPT mode, where the PV generator delivers the
lenging gap in the PMS field is to preserve a smooth transition maximum power to feed the load. In energy surplus, the PV gen-
between different energy sources during several operation modes erator will be redirected to charge the battery if it has not been
(e.g. the battery isolation when it becomes fully charged and fully charged (SOC ≤ SOCmax). Alternatively, the second mode is the
when it approaches a deep discharging state) [14]. LPP mode, where the battery reaches its full charge and the role
Consequently, this desirable behaviour provides a more ex- of the PV generator is restricted in satisfying the load demand
cellent energy balance and increases energy efficiency. However, [11, 12].
in a conventional PMS of hybrid systems, the operating modes To ensure the two operation modes of the PV generator, this
are selected by abruptly switching between them, resulting in generator must be connected to a DC/DC converter, which is
instantaneously varying power references [14]. These kinds of considered an impedance adapter between the DC bus and this
references are not easy to track by most regulators, particularly generator, capable of changing its operating point to follow the
conventional ones like PI (proportional integral) controllers [14]. reference point imposed by the PMS-FLC system.
Besides, they will cause transient regimes, which influence the The PV system control based on PMS-FLC shown in Fig. 2a has
system’s stability and minimize its efficiency. To face this chal- three inputs:
lenge, introducing FLC into the PMS structure seems to be an ef-
fective solution capable of guaranteeing a smooth and flexible • The error (EPV) between the powers requested by the load
switching between its different modes via introducing an inter- and that delivered by the PV generator; this input is used
mediate phase between them. The PMS-FLC is designed and con- to control the system in LPP mode.
figured to scale all expected system scenarios following the below • The ratio between the difference of the power of the PV
convention: generator and the difference in its voltage (∆PPV/∆VPV); this
674 | Clean Energy, 2022, Vol. 6, No. 4

input is used to locate the operating point of the PV gener- There are several methods in the literature to achieve MPPT
ator to control the system in MPPT mode or LPP mode. [21] and the best among them are those that have a variable step
• The SOC of the battery; this input is needed to choose be- (∆VPV_ref) when searching for MPP to give a reduced response time
tween the two modes, LPP or MPPT. Indeed, there are three and low ripple around MPP, and this will increase the efficiency
proposed SOC levels: low SOC is the level where the battery of using a PV generator [21–23]. The MPPT is based on fuzzy logic
reaches the deep discharge; high SOC is the level where the with variable steps (VPV_ref) in this section. This method has sev-
battery is in full charge state; and the medium level, which eral advantages, such as better performance, robustness and
is between the two preceding levels. According to these simplicity [21–23]. Fig. 3 shows the seven triangular MFs adopted
three levels, the PMS-FLC chooses the operating mode of for MPPT inputs and output, the rules base of MPPT indicated
PV as given in Table 1. in Table 2. The inputs EPV, (∆PPV/∆VPV) and the output (∆VPV_ref) are
divided into seven fuzzy sets: NB (negative big), NM (negative me-
As illustrated in Fig. 2b, the three previously mentioned inputs dium), NS (negative small), Z (zero), PS (positive small), PM (posi-
are used by the PMS-FLC to give a single output representing the tive medium) and PB (positive big).
required PV generator output voltage. This reference voltage will
use as the input of the DC/DC converter regulator. 2.1.2 LPP mode

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

This regulator, structured in two cascading loops with FLC con- This mode is carried out when the battery is fully charged
trollers, aims to give the appropriate duty cycle needed to gener- (SOC = high); therefore, the PV generator delivers power equal to
ate the switching signal via the pulse-width modulation block, the power requested by the load. In this case, the regulator uses
as depicted in Fig. 2c, where ICPV_ref and ILPV_ref are the reference the inputs (∆PPV/∆VPV) to determine the current operating point
current of the capacitor CPV and inductor LPV, respectively, accord- of the PV generator and the input EPV to cancel the error. The
ing to the first FLC regulator, ILPV and VLPV are the current and the PMS-FLC uses the rule base mentioned in Table 2 to provide the
voltage of inductor LPV, respectively, and duty_PV is the duty cycle of output VPV_ref. Therefore, it ensures regulation on the side where
the PV generator’s DC/DC converter. (∆PPV/∆VPV) is negative, as illustrated in Fig. 4.
In Fig. 2b, the error between the load and PV powers EPV = PL – PPV,
with PL = ILoad Vdc and PPV = IPV VPV, where Vdc and VPV represent the 2.1.3 PMS-FLC of PV generator
DC-bus voltage and the PV voltage, respectively, and ILoad and IPV As indicated previously, the PMS-FLC applied on the PV ensures
represent the load current and PV current, respectively. The PV its mode of operation and selects the suitable operating mode ac-
power variation is ∆PPV = PPV (K) – PPV (K – 1) and the PV voltage cording to the battery’s SOC, as indicated in Table 2. In this case,
variation is ∆VPV = VPV (K) – VPV (K – 1). The PV reference voltage the SOC input is divided into three fuzzy sets: Z (SOC = low), PS
variation is ∆VPV_ref = VPV_ref (K) – VPV_ref (K – 1), where VPV_ref represents (SOC = medium) and PB (SOC = high). The corresponding MFs are
the PV output reference voltage. illustrated in Fig. 3d. The other two inputs and output remain the
same as defined previously.
2.1.1 MPPT mode
This mode applies as long as the battery does not reach full 2.1.4 PSO-based approach to determining the optimized
charge, in which the PV generator will give its maximum power MF settings and gain values
regardless of the load level. The maximum power point (MPP) is
The performance of the proposed PMS-FLC system is closely re-
located at the top of the power curve of PV. If the PV power is
lated to the right choice of FLC parameters, namely the MF param-
on the left of the MPP where the (∆PPV/∆VPV) is positive, the VPV_ref
eters, for instance, the values (V1, V2, V3) of MF presented in Fig.
must be increased to reach the MPP; otherwise, the VPV_ref must be
3 and the four gains values G1 to G4 shown in Fig. 2b. To handle
reduced.
this large number of unknowns, an efficient method based on the
Table 1: PV mode selection based on SOC levels PSO algorithm is adopted thanks to its simplicity and efficiency in
solving complex optimization problems [24].
SOC = low SOC = medium SOC = high PSO is a heuristic optimization method inspired by the behav-
MPPT mode MPPT mode LPP mode iour of a group of birds or fish; it was developed first by Eberhart
and Kennedy in 1995 [25]. In this algorithm, each particle tries to

A B
Lpv
PL + G1PV
–
D PPV ∆VPV_ref
VPV CPV Vdc Cdc ∆ PPV/∆ VPV G2PV in out G3PV + VPV_ref
+
Ts G4PV
SOC Z–1
PWM
EPV
VPV_ref
∆PPV/∆VPV
PMS-FLC DC/DC Duty_PV
PV regulator

SOC

C VPV_ref +
ICPV_ref
– ILPV_ref VLPV
FLC –
– + FLC ×
VPV IPV + –
VPV + Duty_PV
ILPV –
Vdc

Fig. 2: PV generator control system. (a) Global schema, (b) PMS-FLC PV system, (c) DC/DC regulator.
 Unified fuzzy logic controller and power management for an isolated residential hybrid PV/diesel/battery energy system | 675

find the best solution around its search space by changing its vel- ˆ
ocity according to its current one, best own solution and the best (3)
fitness = |E| tdt
global solution. This change will give it a new position converging
towards the best solution. This operation remains repetitive until The error E in Equation (3) represents the difference between
a criterion is met. In standard PSO, the particles are manipulated the desired powers in MPPT mode or LPP mode and the current
according to the following equations [24]: PV generator power:
Ä ä Ä ä
(1) vit+1 = wvti + c1 r1 pi,d − xti,d + c2 r2 pg,d − xti,d E = Pref − PPV
(4)

xt+1
(2)
i,d
= xti,d + vt+1
i
2.2 Battery control
where vt+1 and vti represent the current and future velocity of The proposed PMS categorizes the battery as a second energy
i
source after the PV generator. It will intervene when the PV gen-
the i-th particle, xt+1
and
i,d
xti,d
represent the current and the fu-
erator cannot provide the requested load power.
ture positions of the i-th particle and w represents a weighting
It will be controlled by regulating the voltage of the DC bus via
function to control the effect of the particle’s current velocity on
a bidirectional DC/DC converter, as presented in Fig. 5a. The DC/
its future velocity. The variable pi,d represents the best solution of

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

DC converter regulator is based on two nested loops: the external
particle i at iteration t, and pi,g represents the best global solution.
loop is responsible for regulating the DC bus voltage (Vdc) and the
The two weighting factors, c1 and c2, determine the importance of
internal one is in charge of regulating the battery current (Ibat). As
the current local and global best solutions. Finally, r1 and r2 are
illustrated in Fig. 5b, the two loops are based on a standard FLC
two random numbers within the range of [0, 1].
controller [27], with the error and its derivative as inputs and the
To achieve the best solutions from PSO, its implementation fol-
reference command as an output.
lows the following steps [20]:
The two inputs, E and ∆ E, and one output of the FLC controller,
illustrated in Fig. 5c, are represented by seven fuzzy variables de-
• Selecting the FLC parameters to be optimized: The param- noted by NB, NM, NS, Z, PS, PM and PB. The fuzzy inference of the
eters to be optimized using PSO are the values V1_EPV, V2_ FLC is found on the rule base mentioned in Table 3.
EPV and V3_ EPV of the MFs of the EPV input; V1_dp/dv, V2_ dp/ The SOC of the battery must remain between two limits: deep
dv and V3_ dp/dv of the ∆PPV/∆VPV input as in Fig. 3b; V1_ discharge (SOCmin) and full charge (SOCmax); both generators (i.e. PV
SOC and V2_SOC of the SOC input; and V1_s, V2_s and V3_s and diesel) participate in achieving this requirement.
of the ∆VPV_ref output as in Fig. 3c with 0 ˂ V1 ˂ V2 ˂ V3 ˂ 1 and
SOC_min ˂ V1_SOC and V2_SOC ˂ SOC_max as in Fig. 3d, as well as 2.3 Diesel generator control
the gains 0 ˂ G1PV ˂ 1 and 0 ˂ G2PV ˂ 1 of two inputs and the The diesel generator runs when the battery SOC reaches its mini-
two gains of the output G3PV, G4PV of Fig. 2b. The correct mum level, while the power generated by the PV generator is in-
search intervals for each value will give a better solution sufficient. In this case, the diesel supplements the PV generator
with fewer iterations. power to cover the necessary load power demand.
• Performing the simulation: The simulation is done to cover The diesel voltage is rectified to give a fixed voltage. This DC
all the foreseen cases of the system’s operation. In this source is also connected to the DC bus via a DC/DC converter,
case, the PV generator works in two modes (LPP and MPPT) which will be controlled by the PMS-FLC as depicted in Fig. 6a.
under the variation of irradiation, temperature and load. The PMS-FLC applied to the diesel generator needs three in-
• Fitness function definition: To design an optimal PMS, puts: Edies, ∆Edies and SOC.
several objective functions can be used [26], for example, Edies = ILdies_ref – ILdies is the diesel error, ILdies is the inductor current
mean absolute of error, integral absolute of error, mean of diesel’s DC/DC converter, ILdies_ref is the inductor reference cur-
square of error, the integral square of error and integral rent, ∆Edies = Edies(K) – Edies (K – 1) is the diesel error variation. The in-
time absolute of error (ITAE). Herein, the ITAE criterion is puts Edies and ∆Edies are used to adjust its diesel power and a third
adopted as the objective function: one (SOC) is used to turn it on or off. The output of PMS-FLC gives

A µ B µ
NB NM NS Z PS PM PB NB NM NS Z PS PM PB

EPV ∆PPV /∆VPV

–V3_EPV –V2_EPV –V1_EPV 0 V1_EPV V2_EPV V3_EPV –V3_dp/dv –V2_dp/dv –V1_dp/dv 0 V1_dp/dv V2_dp/dv V3_dp/dv

C µ D µ
N
NM NS Z PS PM PB
B Z PS PB

∆VPV_ref SOC

–V3_s –V2_s –V1_s 0 V1_s V3_s 0 SOCmin V1_SOC V2_SOC SOCmax

V2_s

Fig. 3: Inputs and output MFs. (a) EPV, (b) ∆ PPV/∆VPV, (c) ∆ VPV_ref, (d) SOC..
676 | Clean Energy, 2022, Vol. 6, No. 4

the duty cycle of the DC/DC converter, as illustrated in Fig. 6b. and Imp represent the MPP voltage and current, respectively, and
The diesel generator is operating in modes according to the SOC all these parameters are set under standard test conditions (STC).
provided in Table 4. The performances test of the proposed PMS-FLC was done in
The rules of starting and stopping modes are indicated in Table two stages. The first stage concerns the performance of the PV
5. generator when it is operating in MPPT mode or LPP mode, and
the second stage is when there is a change in the operating mode
of the system between the three sources (PV, battery and diesel).
3 Results and discussion
To verify the efficiency of the proposed PMS-FLC, the proposed 3.1 PSO optimized PMS-FLC performance of the
system has been simulated on the MATLAB®/SIMULINK® PV generator
(Simscape ElectricalTM) system environment. The number of The PSO is applied under the following conditions:
panels and batteries were chosen to cover the load, which is a
house. The parameters of each block are indicated in Table 6, • The power requested by the load is equal to 2000 W. A step
where Pmax represents the PV maximum power, Voc,n represents the change in solar irradiation from 0 to 1000 W/m2 is per-
open-circuit voltage, Isc,n represents the short-circuit current, Vmp formed at t = 0.1 s under T = 25°C. The simulation duration

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

is 0.5 s with the PV generator operating in MPPT mode
during [0 0.3] s and LPP mode during [0.3 0.5] s.
120 MPP • The reference power is set to Pref = PMPP = 2996.5 W during
[0.1 0.3] s and Pref = PLPP = 2000 W during [0.3 0.5] s as shown
in Fig. 8.
100

LPP Fig. 7 presents the evolution of the fitness function, which

80 reaches its minimum value equal to 0.857 after 100 iterations.
PV power (W)

Table 7 shows the values of the PMS-FLC membership param-

eters of the PV generator indicated in Fig. 3 optimized by PSO.
60 Fig. 8 illustrates the obtained results using optimal param-
eters using the PSO algorithm. These results are compared
with those obtained by the classical method using the per-
40
turb and observe (P&O) method to reach the MPP or LPP as
in [11, 12]. Table 8 gives the performance of each method, in
20 which the proposed method shows better results in both dy-
namic (reduced rise time) and static (reduced ripples) perform-
ances, which leads to the best yield. On the other hand, the
0 P&O method with a higher step (0.1 V) exhibits good dynamic
0 10 20 30 40 50
performances with reduced rise time. However, the static per-
PV tension (V)
formance has deteriorated with the presence of high ripples. By
Fig. 4: Limit mode of PV generator. contrast, the P&O method with a lower step (0.01 V) reduces the

Table 2: Rules base for MPPT mode and LPP mode

Mode MPPT mode (SOC = Z or SOC = PS) LPP mode (SOC = PB)

EPV NB NM NS Z PS PM PB NB NM NS Z PS PM PB
∆PPV/∆VPV

NB NB NB NB NB NB NB NB PB PM PS Z NS NM NB
NM NM NM NM NM NM NM NM PB PM PS Z NS NM NM
NS NS NS NS NS NS NS NS PB PM PS Z NS NS NS
Z Z Z Z Z Z Z Z NB NM NS Z Z Z Z
PS PS PS PS PS PS PS PS PS PS PS PS PS PS PS

Table 3: Rules base for FLC controller

E NB NM NS Z PS PM PB
∆E

NB NB NB NB NB NM NS Z
NM NB NB NB NM NS Z PS
NS NB NB NM NS Z PS PM
Z NB NM NS Z PS PM PB
PS NM NS Z PS PM PB PB
PM NS Z PS PM PB PB PB
PB Z PS PM PB PB PB PB
 Unified fuzzy logic controller and power management for an isolated residential hybrid PV/diesel/battery energy system | 677

A T2
C
Lbat
D2
E GE
T1
D1 Vdc Cdc ∆com Ts
in out Gcom com
Vbat
Z–1
+ ∆E G∆E
Z–1 –
Ibat

Vdc_ref DC/DC Duty_bat

PWM
regulator
Vdc

B
ICdc_ref
Vdc_ref + FLC Pdc_ref
– X × Ibat_ref
Vdc + FLC –
Vdc

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

– ×
Vbat – Vbat + Duty_bat
Ibat Vdc –

Fig. 5: Battery control system. (a) Global schema, (b) DC/DC converter FLC control, (c) FLC schema.

Table 4: Diesel generator operating modes according to SOC

SOC = Z SOC = PS SOC = PB

Starting Stopping Stopping

Table 5: Rule bases of PMS-FLC diesel generator control in starting and stopping modes

Mode Starting (SOC = Z) Stopping (SOC = PS or SOC = PB)

Edies NB NM NS Z PS PM PB NB NM NS Z PS PM PB
∆Edies

NB NB NB NB NB NM NS Z NS NS NS NS NS NS NS
NM NB NB NB NM NS Z PS NS NS NS NS NS NS NS
NS NB NB NM NS Z PS PM NS NS NS NS NS NS NS
Z NB NM NS Z PS PM PB NS NS NS NS NS NS NS
PS NM NS Z PS PM PB PB NS NS NS NS NS NS NS
PM NS Z PS PM PB PB PB NS NS NS NS NS NS NS
PB Z PS PM PB PB PB PB NS NS NS NS NS NS NS

A B
Ldies X IL
EPV dies_ref Edies
+ G1dies
Cdies –
T Vdc Vdc –
D Cdc
+
ILdies Z–1 – G2dies

Diesel Rectifier
PWM
Duty_dies SOC

ILdies

EPV
PMS-FLC
Diesel Vdc
SOC ∆duty_dies Ts
in out G3dies Duty_dies
Z–1

Fig. 6: Diesel control system. (a) Global schema, (b) PMS-FLC diesel controller.
678 | Clean Energy, 2022, Vol. 6, No. 4

Table 6: Simulation parameters of PV/battery/diesel and DC/DC converters

Generator Model DC/DC converter

PV generator Module: SAMSUNG SDI PV-MBA 1BG 250 Boost, LPV = 5 mH,
Pmax = 250 W, Voc,n = 37.9 V, Isc,n = 8.85 A Vmp: 30.3 V, CPV = 700 uF, frequency = 20 kHz
Imp:8.24 A, quantity: 4 in parallel and 3 in series
Battery Model: Lithium-ion voltage = 24 V, Capacity = 100 Ah,
Initial SOC = 50%, quantity: 5 in series
Diesel + rectifier Buck, L = 9 mH, frequency = 20 kHz

Table 7: The parameter values of membership functions optimized using PSO

EPV input ∆PPV/∆VPV input ∆VPV_ref output The gains SOC input

V1_EPV 0.0290 V1_dp/dv 0.3074 V1_s 0.4341 G1PV 0.0001 SOC_max 90

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

V2_EPV 0.4947 V2_dp/dv 0.5401 V2_s 0.5316 G2PV 0.0231 SOC_min 20
V3_EPV 0.8032 V3_dp/dv 0.5985 V3_s 0.7536 G3PV 0.1630 V1_SOC 30
G4PV 0.0535 V2_SOC 80

Best function value: 0.857137

2
Function value

1.5

0 20 40 60 80 100
Iteration
Stop Pause

Fig. 7: Fitness function versus iteration number.

2040
2996
2020
2995
2000

2994 1980
0.3 0.35 0.4 0.45 0.5
0.1 0.15 0.2 0.25 0.3

3000
PV generator power (W)

2000

P&O 0.01v
P&O 0.1v
1000
Proposed FLC

0
0 0.1 0.2 0.3 0.4 0.5
Time (s)

Fig. 8: PV generator power with different methods (P&O and proposed method).

Table 8: Performances of different methods (P&O and proposed method)

MPPT (mode) LPP (mode) Yield (%)

Rise time (s) Ripple (W) Rise time (s) Ripple (W)

P&O (step = 0.01 V) 0.062 0.5 0.013 18 91.8

P&O (step = 0.1 V) 0.008 1.8 0.006 23 99.35
Proposed PMS-FLC 0.004 0.4 0.003 3.1 99.50
 Unified fuzzy logic controller and power management for an isolated residential hybrid PV/diesel/battery energy system | 679

3000
2000
1000

2.4 2.45 2.5 2.55

A B
100
3000 Proposed method

PV power (W)
Classic method
90
2000
SOC (%)

80
1000
70

60 0
0 1 2 2.5 3 4 5 0 1 2 2.5 3 4 5
Time (s) Time (s)

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

C D 1000
202
0
200
198 –1000
196 –2000
Proposed method
194 Classic method –3000
2.45 2.5 2.55 2.55 2.45 2.55
1000
205 Battery power (W)
0
200
Vdc (V)

–1000
–2000
195
Proposed method –3000 Proposed method
190 Classic method
Classic method –4000
0 1 2 2.5 3 4 5
0 1 2 2.5 3 4 5
Time (s)
Time (s)

Fig. 9: Results of the first scenario. (a) SOC (%), (b) PV power, (c) DC-bus voltage, (d) battery power.

steady-state ripples, but the rise time increases in return. This Scenario 2
classical technique has difficulty providing good performance Under zero irradiation, the battery feeds a fixed load and when
in both dynamic and static performances, making the trade-off it reaches its deep discharge state (SOC = SOCmin), the diesel takes
between them a necessity. charge of supplying the same load.
The yield of each method shown in Table 8 is calculated ac- As indicated in Fig. 9a, the battery reaches its full charge state
cording to the following relationship [28]: (SOC = 80%) at t = 2.5 s. At this instant, the PV generator operating
´ Tf mode changes from MPPT (PPV = 2996 W) to LPP (PPV = 200 W), as
 PPV (t) dt
η = ´ T0
T
100 (5) illustrated in Fig. 9b. This change generates a sudden disturbance
f
T0 Pref (t) dt
at the level of the DC-bus voltage, as depicted in Fig. 9c. Fig. 9d
confirms that the battery is fully charged since its power passes
where To = 0.1 s, Tf = 0.5 s and Pref and PPV are the power reference from –2696 to 0 W.
and the PV generator power. Classical PMS based on abrupt mode change leads to power
oscillations and DC-bus voltage spikes. These drawbacks are
3.2 Performance of the PMS-FLC of the global strongly lowered when the fuzzy-logic-based PMS is applied.
system when it changes its operating modes Indeed, the PMS-FLC can decrease the DC-bus voltage (attenu-
The proposed PMS-FLC will be compared with the classical ation of 2%) and battery power perturbations and reduce the
method adopted in several papers [11, 12] to manage multi- power time responses significantly. This is due to the soft mode
source hybrid systems. The comparison is made in the following change anticipated by the proposed PMS using the SOC input
two scenarios: given in Fig. 3d.
Fig. 10 represents the system responses when performing the
Scenario 1 second scenario. The battery starts by feeding alone an isolated
Under maximum irradiation (1000 W/m2), the PV generator, load of 1500 W, as shown in Fig. 10a, and the battery reaches
operating in MPPT mode, supplies the load and charges the bat- its deep discharge state (SOC = 20%) at time t = 3 s. This can be
tery simultaneously. When the battery reaches its full charge manifested by a battery power change from 1500 to 0 W, as pre-
(SOC = SOCmax), the PV generator switches from MPPT mode to LPP sented in Fig. 10d. Simultaneously, the diesel power goes from 0 to
mode. 1500 W, as shown in Fig. 10b, satisfying the load need. Once more,
680 | Clean Energy, 2022, Vol. 6, No. 4

1500

1000

500

0
2.75 2.8 2.85 2.9 2.95 3
A B
60 3000
Proposed method

Diesel power (W)

Classic method
40 2000
SOC (%)

20 1000

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

0
0 1 2 3 4 5 0 1 2 3 4 5
Time (s) Time (s)
C D
1500
206
1000
203 500
201 0
200
–500
198
2.8 2.9 3 2.8 2.9 3 3.1
3000
Proposed method
Battery power (W)

205 Classic method

2000
Vdc (V)

200
1000
195 Proposed method 0
Classic method
190
0 1 2 3 4 5 –1000
0 1 2 3 4 5
Time (s) Time (s)

Fig. 10: Results of the second scenario. (a) SOC (%), (b) diesel power, (c) DC-bus voltage, (d) battery power.

the proposed PMS based on a fuzzy system ensures the same Conflict of interest statement
power balance as the classical one, but with a gradual increase in
We wish to confirm that there are no known conflicts of interest
the diesel generator power in Fig. 10b.
associated with this publication and there has been no signifi-
This smooth transition is achieved through inserting an inter-
cant financial support for this work that could have influenced
mediate phase controlled by the SOC input, whose MFs are given
its outcome.
in Fig. 3d, which results in fewer disturbances on the DC-bus volt-
age (0.5%), as illustrated in Fig. 10c.

References
4 Conclusion [1] Khelif A, Talha A, Belhamel M, et al. Feasibility study of hy-
This paper proposes an optimal fuzzy-logic-based power man- brid diesel-PV power plants in the southern of Algeria: case
agement strategy to maximize the efficiency of a hybrid system study on AFRA power plant. Int J Electr Power Energy Syst, 2012,
combining two energy sources (PV and diesel) and a battery stor- 43:546–553.
age system connected to the same DC bus and feeding an isolated [2] Halabi LM, Mekhilef S, Olatomiwa L, et al. Performance ana-
load. The simulation of PMS-FLC for the different operating con- lysis of hybrid PV/diesel/battery system using HOMER: a
ditions of the hybrid system shows excellent stability of the DC case study Sabah, Malaysia. Energy Convers Manag, 2017,
bus and a low ripple of the power of the sources in the transient 144:322–339.
regime, which reflects the advantage of the PMS-FLC proposed [3] Yu W, Li S, Zhu Y, et al. Management and distribution strat-
over the conventional PMS using the straightforward transition, egies for dynamic power in a ship’s micro-grid system based on
and this is due to the smooth change between PV and diesel gen- photovoltaic cell, diesel generator, and lithium battery. Energies,
erators while keeping the SOC of the batteries between 20% and 2019, 12:45051–45018.
80%. This increases the performance with overshoot attenuation [4] Shezan SA. Feasibility analysis of an islanded hybrid wind-
at the DC-bus voltage of 2% when changing the mode and an im- diesel-battery microgrid with voltage and power response for
provement in the PV generator efficiency by 99.5%. offshore Islands. J Clean Prod, 2021, 288:125568.
 Unified fuzzy logic controller and power management for an isolated residential hybrid PV/diesel/battery energy system | 681

[5] Ishraque MF, Shezan SA, Rashid MM, et al. Techno-economic solar-wind-storage energy system. Energy Convers Manag, 2017,
and power system optimization of a renewable rich islanded 148:238–250.
microgrid considering different dispatch strategies. IEEE Access, [17] Roumila Z, Rekioua D, Rekioua T. Energy management based
2021, 9:77325–77340. fuzzy logic controller of hybrid system wind/photovol-
[6] Upadhyay S, Sharma MP. Selection of a suitable energy manage- taic/diesel with storage battery. Int J Hydrogen Energy, 2017,
ment strategy for a hybrid energy system in a remote rural area 42:19525–19535.
of India. Energy. 2016, 94:352–366. [18] Cabrane Z, Ouassaid M, Maaroufi M. Battery and supercapacitor
[7] Aziz AS, Tajuddin MFN, Adzman MR, et al. Energy management for photovoltaic energy storage: a fuzzy logic management. IET
and optimization of a PV/diesel/battery hybrid energy system Renew Power Gener, 2017, 11:1157–1165.
using a combined dispatch strategy. Sustain, 2019, 11:683. [19] Sinha S, Bajpai P. Power management of hybrid energy storage
https://doi.org/10.3390/su11030683. system in a standalone DC microgrid. J Energy Storage, 2020,
[8] Kartite J, Cherkaoui M. Study of the different structures of hy- 30:101523.
brid systems in renewable energies: a review. Energy Procedia, [20] Cheng PC, Peng BR, Liu YH, et al. Optimization of a fuzzy-logic-
2019, 157:323–330. control-based MPPT algorithm using the particle swarm optimi-
[9] Olatomiwa L, Mekhilef S, Ismail MS, et al. Energy management zation technique. Energies, 2015, 8:5338–5360.

Downloaded from https://academic.oup.com/ce/article/6/4/671/6672708 by guest on 22 January 2024

strategies in hybrid renewable energy systems: a review. Renew [21] Rezk H, Eltamaly AM. A comprehensive comparison of different
Sustain Energy Rev, 2016, 62:821–835. MPPT techniques for photovoltaic systems. Sol Energy, 2015,
[10] Belila A, Benbouzid M, Berkouk EM, et al. On energy manage- 112:1–11.
ment control of a PV-diesel-ESS based microgrid in a stand-a- [22] Macaulay J, Zhou Z. A fuzzy logical-based variable step size P&O
lone context. Energies, 2018, 11:21641–21622. MPPT algorithm for photovoltaic system. Energies, 2018, 11:1340.
[11] Choudar A, Boukhetala D, Barkat S, et al. A local energy manage- https://doi.org/10.3390/en11061340.
ment of a hybrid PV-storage based distributed generation for [23] Chen YT, Lai ZH, Liang RH. A novel auto-scaling variable step-
microgrids. Energy Convers Manag, 2015, 90:21–33. size MPPT method for a PV system. Sol Energy, 2014, 102:247–256.
[12] Guichi A, Talha A, Berkouk EM, et al. Energy management and [24] Amer M, Namaane A, M’Sirdi NK. Optimization of hybrid renew-
performance evaluation of grid connected PV-battery hybrid able energy systems (HRES) using PSO for cost reduction. Energy
system with inherent control scheme. Sustain Cities Soc, 2018, Procedia, 2013, 42:318–327.
41:490–504. [25] Kennedy J, Eberhart R. Particle swarm optimization. In: Proc IEEE
[13] Yi Z, Dong W, Etemadi AH. A unified control and power manage- Int Conf neural networks, Perth, WA, Australia, 27 November–1
ment scheme for PV-Battery-based hybrid microgrids for both December 1995, 1942–1948.
grid-connected and islanded modes. IEEE Trans Smart Grid, 2018, [26] Noshadi A, Shi J, Lee WS, et al. Optimal PID-type fuzzy logic con-
9:5975–5985. troller for a multi-input multi-output active magnetic bearing
[14] Ji X-P, Liu Q-J, Liu Z, et al. Coordinated control and power man- system. Neural Comput Appl, 2016, 27:2031–2046.
agement of diesel-PV-battery in hybrid stand-alone microgrid [27] Douiri MR, Cherkaoui M, Essadki A. Genetic algorithms based
system. J Eng, 2019, 2019:5245–5249. fuzzy speed controllers for indirect field oriented control of
[15] Koohi-Kamali S, Rahim NA, Mokhlis H. Smart power manage- induction motor drive. Int J Circuits, Syst Signal Process, 2012,
ment algorithm in microgrid consisting of photovoltaic, diesel, 6:21–28.
and battery storage plants considering variations in sunlight, [28] Kermadi M, Salam Z, Ahmed J, et al. A high-performance
temperature, and load. Energy Convers Manag, 2014, 84:562–582. global maximum power point tracker of PV system for rapidly
[16] Derrouazin A, Aillerie M, Mekkakia-Maaza N, et al. Multi in- changing partial shading conditions. IEEE Trans Ind Electron, 2020,
put-output fuzzy logic smart controller for a residential hybrid 68:2236–2245.