IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 9, NO.

2, APRIL 2018 853

Load Frequency Control in Microgrids Based

on a Stochastic Noninteger Controller
Mohammad-Hassan Khooban , Member, IEEE, Taher Niknam , Member, IEEE, Mokhtar Shasadeghi,
Tomislav Dragicevic , Senior Member, IEEE, and Frede Blaabjerg , Fellow, IEEE

Abstract—In this paper, an adaptive multiobjective fractional- wind turbine generator (WTG) systems [1]–[3]. In this context,
order fuzzy proportional–integral–derivative controller is pro- the local harness of RES through distributed generators (DGs)
posed for the load frequency control (LFC) of islanded microgrids is quite significant as it can decrease power loss in transmission
(MGs), while benefiting from the assets of electric vehicles (EVs)
in this respect. Although the use of battery energy storage systems lines and increase the overall reliability of the electrical power
(BESS) can solve the unbalance effects between the load and supply systems [4]. The concept of the microgrid (MG) has been pro-
of an isolated MG, their high cost and tendency toward degrada- posed to locally co-optimize the operation of a number of DGs
tion are restrictive factors, which call for the use of alternative and to further enhance the controllability, reliability, flexibil-
power balancing options. In recent years, the concept of utilizing ity and power quality of the overall electrical systems, while
the BESSs of EVs, also known as vehicle-to-grid (V2G) concept, for
frequency support of MGs has attracted a lot of attention. In order postponing the construction of new power stations at the same
to allow the V2G controller operate optimally under a wide range time [5].
of operation conditions caused by the intermittent behavior of re- From the utility point of view, MG is an independent unit
newable energy resources, a new multiobjective fractional-order which can be operated in both grid-connected and isolated
control strategy for the EVs in V2G scenarios is proposed in this modes [6]. In the grid connected mode, the main target is usu-
paper. Moreover, since the performance of the controller depends
on its parameters, optimization of these parameters can play a ally to perform auxiliary services to the overhead system and
significant role in promoting the output performance of the LFC optimize the internal energy management at the same time [7].
control; hence, a modified black hole optimization algorithm is uti- On the other hand, the isolated operation of MG can be used for
lized for the adaptive tuning of the noninteger fuzzy PID controller supplying remote applications (telecom, rural households, etc.)
coefficients. The performance of the proposed LFC is evaluated by or for enhancing the power quality of weak grids by manually
using real world wind and solar radiation data. Finally, the exten-
sive studies and hardware-in-the-loop simulations are presented disconnecting it from the overhead system. In the latter mode,
to prove that the proposed controller tracks frequency with lower due to the low inertia of the MGs compared with the grid, it is
deviation and fluctuation and is more robust in comparison with a challenge to maintain the active and reactive power balances
the prior-art controllers used in all the case studies. between the supply and consumption, especially in the presence
Index Terms—Load frequency control (LFC), modified black of intermittent RES and frequent load variations [8], [9]. The
hole algorithm (MBHA), microgrid (MG), fractional controller, most common solution to this problem in isolated MGs was
electric-vehicle (EV). the installation of backup systems such as micro turbines and
energy storage systems (ESSs).
Besides RES, electric vehicles (EVs) and plug-in hybrid elec-
I. INTRODUCTION
tric vehicle (PHEV) technologies are other key technologies
N recent years, due to the decline of conventional fossil
I fuel resources and the environmental pollutions they caused,
modern electrical power systems face increased deployment of
that will lead to a carbon-neutral society [10]. Since EVs are
environmentally friendly and can significantly reduce the use
of gasoline, it is predicted that they will dominate the future
renewable energy sources (RESs) like photovoltaic (PV) and of automotive industry. In the PHEV technology, the bidirec-
tional flow of power between the grid and the vehicle’s battery
can be achieved, so that its remaining capacity can be used for
Manuscript received October 18, 2016; revised February 17, 2017 and March
31, 2017; accepted April 21, 2017. Date of publication October 16, 2017; date of ancillary services such as peak shaving or frequency support.
current version March 20, 2018. This work was supported by the Iran National However, the charging and discharging of PHEV batteries can
Science Foundation under grant to collect data. Paper no. TSTE-00794-2016. increase the complexity of the overall frequency control of the
(Corresponding author: Mohammad Hassan Khooban.)
M. H. Khooban, T. Niknam, and M. Shasadeghi are with the Department system [11], [12].
of Electrical and Electronic Engineering, Shiraz University of Technology, When considering the autonomous MG, both RES and EVs
Shiraz 13876-71557, Iran (e-mail: [email protected]; [email protected]. need to be coordinated together to limit the frequency fluctu-
ir; [email protected]).
T. Dragicevic and F. Blaabjerg are with the Department of Energy Technol- ation by compensating for the mismatch between generation
ogy, Aalborg University, DK-9220 Aalborg, Denmark (e-mail: [email protected]; and demand. This functionality is often referred to as load fre-
[email protected]). quency control (LFC). The proper LFC can not only guarantee
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. the frequency stability of the MG but also increase its efficiency
Digital Object Identifier 10.1109/TSTE.2017.2763607 (fuel saving). In order to improve the response of LFC, many

1949-3029 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 854 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 9, NO. 2, APRIL 2018

controllers including conventional PID control [13], intelligent

control [14], adaptive control [12], robust control [15] and MPC
control [11] have been applied to the DGs of islanded MGs.
In [16], an MPC-based coordinated control of the blade pitch
angles of the WTG and PHEV has been proposed for the LFC.
The MPC controller is applied in order to smooth wind power
production and decrease the number of required PHEVs. In [17],
a new method is suggested for the LFC in MGs by using an in-
telligent PI controller for improving the robustness of the whole
of system. Moreover, the integral square error (ISE) is used for
optimum tuning of PI’s parameters to increase the performance
of the proposed controller [17].
Since the operation conditions of the LFC can widely change,
the PI controller which is tuned for nominal conditions cannot Fig. 1. General scheme of Micro-Grids.
work properly in other conditions. So, in order to solve this
problem, the fuzzy logic (FL) which adjusts the control param-
eters according to the operating point is proposed in [11]. One
drawback of the system is that its good performance is achieved
in only some specific member functions. In [18], the applica-
tion of robust H-infinity control in the LFC of an isolated MG
is studied. The control method, which has been suggested in
[18], is very complicated and hence not attractive for real world
applications. As a result, the small signal analysis of an isolated
MG in the presence of energy storage unit has been proposed in
[19]. Besides, in order to increase the robustness of the isolated
MG, the hierarchical control is presented in [20]. Fig. 2. The electric vehicle model in Microgrid LFC.
In order to overcome the aforementioned difficulties, this
paper introduces a new adaptive approach by using a novel The power grid and the MG operation are controlled by the
optimization technique for the adaptive tuning of the most distribution management system (DMS) and the MG dispatch
common fractional-order fuzzy PID (FOFPID) controller for system (MGDS), respectively. Also, the bidirectional informa-
the EV that supports the LFC in islanded MG(s). FOFPID tion transfer can be achieved through communication links [11].
parameters are tuned automatically according to the online
measurements, by applying MBHA. Unlike the classical tun- B. Electric Vehicle Model
ing methods which are not suitable for providing a useful
Generally, there are different numbers of EVs in each EV
performance over a wide range of operation conditions, many
station. As a result, using an equivalent EV with a different in-
advantages are offered by the proposed optimal tuning scheme
verter capacity can handle the modeling of the EVs’ behavior.
for the frequency control of an MG with many DGs and RESs.
The equivalent EV model, proposed in [11], [23] for the LFC
Moreover, the proposed method is significantly less complex
study, is shown in Fig. 2. In this figure, Te and Δue represent the
in comparison to the above-mentioned approaches, making
time constant and the LFC command signal to the EV, respec-
it more convenient for practical applications. The simulation
tively. The interfacing inverter capacity limits are shown by ±μe
study is performed on a complex MG, including different loads
and ±δe which represent the ramp rate limits. The lower and
and RESs to demonstrate the effectiveness of the proposed
upper limits of the EV’s controllable energy are represented by
control scheme; the superiority of the suggested controller over
Em in and Em ax , correspondingly. The charging or discharging
Multi-Objective Proportional–Integral–Derivative (MOPID),
power of the EV is represented by ΔPE . The positive sign of
Multi-Objective Fuzzy Proportional–Integral (MOFPI) and
ΔPE indicates that the EV is in the discharging mode while the
Multi-Objective Interval Type-2 Fuzzy Logic (MOIT2FLC)
negative sign shows that EV is in the charging mode. ΔPE = 0
controllers is demonstrated in Section VI. To further corroborate
expresses the idle operation mode of the EV.
the performance and robustness of the proposed control system,
experimental validation using the Hardware-in-the-Loop (HiL)
simulation is also given in this paper. C. The Diesel Power System Model
Due to the advantages of the diesel power system, including
II. MODELING OF AN ISOLATED MICROGRID its fast starting speed, low maintenance and high efficiency,
it has been a good backup option in isolated microgrids. The
A. The Model of an Isolated Microgrid controllable DG can instantaneously track load demand changes
Fig. 1 depicts an isolated MG in which DGs such as PVs and through the power control mechanisms it employs [17]. The
WTGs, and energy storage units such as BESS and FESS supply fluctuation of uncontrollable DGs such as WTG, PV and loads
the distributed loads [22]. can be effectively compensated for by the diesel power system.

Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 KHOOBAN et al.: LFC IN MICROGRIDS BASED ON A STOCHASTIC NONINTEGER CONTROLLER 855

Fig. 3. Diesel Power System Model. Fig. 5. Case Study model for the LFC in MG system.

TABLE I
MG POWER SYSTEM’S PARAMETERS

Symbol and Values Symbol and Values

Abbreviation Abbreviation

Tg 2.000 s T AE 0.5 s
Te 1.000 s TP V 1.8 s
R 3.000 TW T G 2s
δe 0.010 KD E G 1/300
μe 0.025 KP V 1
Em a x 0.950 KW T G 1
Em in 0.800 KAE 1/500
D 0.012 Kn 0.6
2H 0.200 KF E S S −1/100
T F ESS 0.100 s KB E S S −1/300
T BESS 0.100 s KF C 1/100
Fig. 4. The overall microgrid scheme including EV for LFC. TF C 4.000 s ‘s’ means second

Fig. 3 shows the transfer function of the diesel power system

However, power loss is inevitable in them due to the bound-
which represents the relationship between the LFC signal and
ary and external contact with the series resistance. Moreover,
the output power of DG. As shown in Fig. 3, the model consists
the generated power of the PV is intermittent and depends on
of first order inertia models of the governor and diesel generator
sun irradiance and temperature; hence, in simulations a random
and two saturation blocks.
power source can be used to model the behavior of PV [13],
In this figure, Δf and ΔuDG represent frequency deviation
[22]. The PV model which is considered as disturbance for the
and the LFC command signal from the DG, respectively. Tg
LFC is depicted in Fig. 4.
and Td denote the time constant of the governor and the diesel
generator, correspondingly. ΔXG shows the condition of the
governor’s valve. The speed regulation coefficient of the DG is F. General Scheme of Microgrid With LFC Controller
shown by R in this Fig. ±μdg and ±δdg represent the power The framework of the proposed LFC and an isolated MG
increment and ramp rate limits, respectively. The output power which consists of EVs, DGs such as PV, WT and diesel gen-
increment of the diesel power system is represented by ΔPDG . erator, energy storage units like BESS and FESS and loads is
ΔPDG = 0 means that demand and generation are in a balance depicted in Fig. 5. As shown in this figure, the PV, Fuel Cell
condition and there is not any need for changing the power. (FC), BESS and FESS units are connected to the AC MG via
ΔPDG > 0 means that the required power is higher than the DC/AC interfacing inverters. The FC is modeled by a third or-
actual power while the ΔPDG < 0 represents the condition that der transfer function [22], [24]. All small scale DGs and energy
the actual power is less than the demand [23]. storage units are connected to the AC bus via a circuit breaker.
The spinning reserve for the secondary frequency control is
D. Model of Wind Turbine provided by the diesel power system.
The parameters of the MG system which is depicted in Fig. 4
Because of the intermittent behavior of wind, the output
are listed in Table I.
power of wind turbines changes according to the different speed
and directions wind takes [13], [22]. By controlling the DG
units and EVs, the uncontrolled WTG can be modeled as a dis- III. THE STRUCTURE OF FRACTIONAL ORDER FUZZY
turbance and a power fluctuation source in LFC. The dynamic PID CONTROLLER
model of WTG is shown in Fig. 4. A. Fractional-Order Calculus (FOC) in Control Systems
Fractional calculus is one of the most important branches
E. Photovoltaic (PV) Generation
of calculus in which the power of the differential and inte-
PV cells which are made from semiconductor materials can gration operators can take a non–integer value. During the
directly convert the energy of photons to electrical energy. last few decades, the FOC has been applied in many fields of
Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 856 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 9, NO. 2, APRIL 2018

Small, Negative Medium, Negative Large, Zero, Positive Small,

Positive Medium and Positive Large. The categorization is based
on triangular MF(s).
Moreover, the former parts of each rule will be made by using
AND function (with hermeneutics of minimum). Besides, the
Fig. 6. The general scheme of Fractional-order PID (FOPID) controller. Mamdani fuzzy inference engine is also applied in this study
[29]. In fuzzy control systems, changing the scaling factors of
TABLE II the fuzzy output has a more positive influence on the perfor-
THE MOFOFPID CONTROLLER RULES SET
mance of the FOFPID controller than does making alterations
in the form of the MF(s). As a result, the tuning of all coefficients
in the FOFPID cannot be equally effective in affecting the total
performance of the control system. The goal of this study is to
explore the result of tuning fractional-orders (δ and λ) since
keeping the rules set and the form of MFs unaltered can notice-
ably increase the total closed loop performance of the FOFPID
controller. Consequently, this method can be more effective in
automatic control systems [25]. Fractional Calculus-Based Con- reducing the volume of calculations, and as a result, the out-
trol Systems can be illustrated by αDtr fx (t) where r ∈ R is the put of the FOFPID controller can be extremely influenced by
order of operation while t and a, are the upper and lower tuning fractional-orders instead of tuning the MF(s) variables or
limits, respectively [26]. There exist a large number of defini- other fuzzy inferencing parameters. So, the next section presents
tions, such as Grünwald–Letnikov (GL), Cauchy integral for- a new modified multi-objective optimization algorithm to turn
mula and Riemann–Liouville (RL), which are applied to define the parameters of proposed controller.
the FOC. However, in automatic control systems, the Cauchy
integral formula is frequently applied for the purpose of realiz- IV. OVERVIEW OF THE ORIGINAL BLACK HOLE
ing the fractional-order differentiations and integrations of the ALGORITHM (BHA)
FO-Fuzzy-PID (FOFPID) controller [27].
 t The idea of BHA was taken from the intelligent collective
1 Dm fx (t) behavior of the stars around a black hole. The movement char-
0Dtr fx (t) = dτ , r ∈ R+ , m ∈
Γ (m − r) 0 (t − τ )r +1−m acteristic of the stars can be expressed as in [30] and [31]:
 
Z+ and m−1≤r <m (1) Xiter iter
m , new = Xm + rand (.) Best
iter
− Xiter
m ;
m = 1, . . . , NPop (3)
B. Design of Fractional Order Proportional
Integral-Derivative (FOPID) Controller where Xiter iter
m and Xm , new represent the position of the target and
the updated agent in iteration iter, respectively. A comprehensive
The main advantage of the FOFPID controller which has introduction with details of this algorithm is provided in [31].
been used in this study is that it encompasses the benefits of
the combination of fuzzy logic with PID controller which is A. IBH Mechanism [31]
comprehensively discussed in [25]. In the proposed method,
Kd and Ke are considered as the fuzzy logic input and, KP I A novel method is suggested for the collapsing process so that
and KP D as fuzzy logic outputs. The advantages of this kind the exploration of the properties of the BHA will be enhanced.
of FOFPID structure over the Model Predictive Control (MPC) Primarily, the updating mechanism of the BHA will be enhanced
and the conventional PID are presented in [28]. In the original through (4):
FPID controller, the power of the input error derivative is inte-  
Xiter iter
m , new = Xm + rand1 (.) Best
iter
− Xiter
m
ger. In this study, in contrast, this power is fractional-order (δ).  
Moreover, the order of the integral in the output is changed by + rand2 (.) Xiter
r − Xiter
m (4)
the fractional order counterpart (μ). The general scheme of the
where r can be regarded as a random number within the range
proposed method is shown in Fig. 6.
of [1, NPop] and r = m. In the second step, the modification
By considering the general scheme of Fig. 6, the control law
of the event horizon of the BH (Riter ) based on the distribution
can be written as:
and collection of the stars should be done using (5) and (6).
UFOFPID = U1 + U2 (2)
m , M ean = Xm , new − Mean
Riter iter iter
; m = 1, . . . , NPop (5)
The other conventional controllers such as the integer PID and N P o p Riter
m , M ean
integer Fuzzy PID can be obtained by considering the values of λ Riter = 0.1 × (6)
and δ as one. Table II displays the set of optimal FOFPID rules.
m =1 NPop
The fuzzy linguistic variables which are similar to the input Moreover, the new concept of Absorption Capacity (AC) is
and output variables are classified as (NS), (NM), (NL), (ZR), proposed for the BH. The elimination of the hyper dispersion of
(PS), (PM), and (PL), referring correspondingly to Negative the solution and the control of the number of the stars located

Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 KHOOBAN et al.: LFC IN MICROGRIDS BASED ON A STOCHASTIC NONINTEGER CONTROLLER 857

in event horizon is achieved in this method. In BHA, after a star

collapses in the BH, a new star is born randomly. The optimum
usage of the data obtained by the members of the population is
achieved by the new formulation expressed in (7).
max Riter
m , Best
m
Xiter
m , new = Best iter
+ . (2 × rand (1, N) − 1)
N
(7)

V. MULTI-OBJECTIVE INTERACTIVE FUZZY SATISFYING Fig. 7. The general scheme of the suggested controller for LFC.
(IFS) METHOD
The value and the number of control signals are among the by the trapezoidal fuzzy membership as represented below:
most important factors to consider in designing and implement- μfw (X) =
ing hardware. In other words, in an application with large val- ⎧
ues of control signals, a big size actuator is required, which ⎪
⎪ 1 for fw (X)  fwm in
⎨
correspondingly increases the total cost of the system. Conse- m ax
f w −f w (X )
for fwm in  fw (X)  fwm ax
quently, effective control of a system requires a smooth control ⎪
⎪
f wm a x −f wm i n
⎩
that can track the set point fast. Using the multi-objective opti- 0 for fw (X)  fwm ax
mization principle can overcome complexity in optimal control (10)
problem. Weighting the cost function of the optimum problem It should be noted that a heuristic algorithm like the MBHA
with some indexes such as the Squared Deviation of Controller merely requires information about the fitness function. As a
Output (ISDCO), Integral of Absolute Error (IAE), Integral of consequence, in order for the online tuning of the FOFPID con-
Time multiplied Squared Error (ITSE) and the Integral of Time troller, the objective functions vector that should be minimized
Multiplied Absolute Error (ITAE) can prevent the high control in the Interactive Fuzzy Satisfying Method can be proposed in
signal. It should be mentioned that the aforementioned objec- terms of the following equation:
tives (IAE, ISDCO, ITSE and ITAE) can act in contrast with
each other. Consequently, formulating the control system as a Min F (X) = [f1 (X) , f2 (X)]T where
multi-objective stochastic optimization problem can reduce the  ∞
value of the control effort. f1 = IT SESet−Point = te2set−point (t) .dt (11)
0
Using the MG system can inherently emulate some of the  ∞
uncertainties of the system pertaining, for instance to its output f2 = ISDCO = Δu2 (t) .dt (12)
variations, and storage batteries; hence, using integer and non- 0
integer controllers can be a good choice for a system in which In this equation, the first objective function is used for provid-
the operation point is widely subject to change. To solve the op- ing the fast tracking of the desired set-point while the accuracy
timization problem, in this paper a new online multi-objective of tracking the set point is achieved by the second objective
optimization method for the optimum tuning of the common function. A schematic representation of the online tuning of
FOFPID controller is proposed. In this method, using online the FOFPID controller based on the MOMBHA technique is
measurement, the Multi-objective MBHA (MOMBHA) is ap- depicted in Fig. 7.
plied for the optimum tuning of the FOFPID controller. The
MOMBHA problem can be formulated by (8). VI. SIMULATION RESULTS
T
Min F (X) = [f1 (X) , f2 (X) , f3 (X) , . . . , fn (X)] In this section, in a trial run to examine the proposed con-
s.t h (X) < 0 trol method, the MG which is depicted in Fig. 4 is simulated
in MATLAB/Simulink software. The parameters of the isolated
g (X) = 0 (8) microgrid are listed in Table I. The performance of the pro-
In this paper, an IFS approach is used to resolve the afore- posed controller is compared with that of the MOPID, MOFPI
mentioned multi-objective problem [32]. In this method, the and MOIT2FLC controllers. Since the respond of the system
sufficient degree of each objective can be defined by the op- depends on the parameters of these controllers, all of the param-
erator. Generally speaking, the non-inferior solution set in the eters are optimized by the MOMBHA optimization algorithm
most optimal solution would be chosen in such a way that the in different controllers. In order to evaluate the performance of
preferences of the operators are satisfied: the proposed control method in the context of the MG depicted
   in Fig. 4, the Hardware-In-the Loop (HIL) simulation approach
F (X) = min maxw ∈W μref w (X) − μw (X)
f  (9) is utilized. The real time HIL method is used to emulate errors
x∈Ω
and delays that do not exist in the classical off-line simulations.
ref
Here, μ is the satisfying degree of the related target, deter- Fig. 8 sketches the HIL setup, consisting of: 1) OPAL-RT as a
mined by the operator in the range of [0, 1]. μfw , on the other Real Time Simulator (RTS) which simulates the MG depicted
hand, denotes the MF value of the wth objective and is designed in Fig. 4; 2) a PC as the command station (programming host)

Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 858 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 9, NO. 2, APRIL 2018

Fig. 8. The Real time experimental setup.

Fig. 10. Frequency response according to the power fluctuation of WPG and
PV.

Fig. 11. Step changes of the load in the interval of 90 seconds.

Fig. 9. Power fluctuation, a- WPG, b- PV.

in which the Matlab/Simulink based code that is executed on

the OPAL-RT is generated; and 3) a router used as a connector
of all the setup devices in the same sub-network. The OPAL-RT
is also connected to the DK60 board through Ethernet ports.
More details about the components of this setup can be found
in [33], [34].
Fig. 12. Frequency response according to the power fluctuation of WPG, PV
and load disturbances.
A. Case 1
In the first step, it is assumed that the load demand in the iso- B. Case 2
lated MG is constant i.e., Δ PL = 0. In other words, the power
In this section, multi-step load variation is applied to the
fluctuation of WTG (ΔPw ) and PVG (ΔPpv ) is consistent in the
LFC as a disturbance. The load steps are depicted in Fig. 11,
LFC system. The wind power fluctuation data extracted from an
while the frequency deviation responds of MOPID, MOFPID,
offshore wind farm in Sweden [35] is depicted in Fig. 9(a), while
MOIT2FLC and MOFOFPID controls are depicted in Fig. 12.
Fig. 9(b) shows the solar radiation data in Aberdeen (United
As displayed in Fig. 12, the frequency deviation overshoot
Kingdom) [36], which was also used in this case study. Fig. 10
is decreased and the LFC control can eliminate the effect of
depicts the frequency response of the simulated system.
load disturbance more effectively when the proposed method,
As shown in Fig. 10, employing MOFOFPID reduces the
rather than the others, is employed. The performances of these
peak value of frequency deviation and this reduction is achieved
controllers are further examined by creating a severe condition,
in the least amount of time as compared to the time when the
applying a large load step at t = 60 s of the simulation. As
MOPID, MOFPI and MOIT2FLC controllers are used. As a
illustrated in Fig. 12, the performance of the LFC in eliminating
result, MOFOFPID can considerably enhance the life of both
the load disturbance can be improved when using the proposed
the batteries and the equipment used in the DG by providing
controller; moreover, as shown in this figure, the settling time
a faster stabilization in the output frequency of the MG while
of the frequency response is decreased significantly.
requiring less frequency adjustment..
Moreover, the simulation results in this case suggest that
C. Case 3
using the MOFOFPID results in the LFC tracking the reference
frequency with fewer overshoots and much smaller settling time To evaluate the robustness of the proposed controller, some
in comparison to the other three controllers. parameters of the isolated MG are changed in the third case

Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 KHOOBAN et al.: LFC IN MICROGRIDS BASED ON A STOCHASTIC NONINTEGER CONTROLLER 859

TABLE III
UNCERTAIN PARAMETERS OF THE MG SYSTEM

Parameters Variation Range

R +15%
D −25%
H +30%
Te −25%
Tg +30%
T F ESS −20%
T BESS +30%

Fig. 14. The frequency deviation of the micro-grid according to Scenario 4.

TABLE IV
THE PERFORMANCE ANALYSIS OF CONTROLLERS

Criteria MOPID MOFPI MOIT2FLC MOFFOPID

SSE 10.8713 8.9211 1.8758 0.6919

MAE 0.0799 0.0327 0.0055 0.0013
MSE 7.6311e-4 2.3527e-4 5.9094e-5 4.9006e-5
Fig. 13. The frequency deviation of the micro-grid according to Scenario 3.

study. The changing of the parameters is made in a scenario structures especially as the overshoots are concerned. Indeed, as
which is displayed in Table III. it can be seen, despite the EV battery restriction that occurs in
As shown in this table, a more severe changing of the pa- the MG in this scenario, the performance of the suggested con-
rameters is applied to evaluate the robustness of the proposed troller is superior to that of the other three that appear in Fig. 14.
control method. Fig. 13 depicts the response of the LFC in the Further noticeable in this figure are the results pertaining to the
scenario by using the proposed as well as the MOPID, MOFPI robustness of the proposed controller against both parameter
and MOIT2FLC controllers. changes and the restriction that was imposed on the system by
As revealed in Fig. 13, the proposed controller enhances the disconnecting the EV. As it can be noticed, none of the MOPID,
performance of the LFC more effectively than the other three MOFPI and MOIT2FLC controllers has such an agreeable per-
control methods, especially when the overshoots are concerned. formance so far as their frequency response against the severe
In other words, the results suggest that the proposed controller changes of parameters (Scenario 3) and their control of the MG
is more robust against the changes in parameters as compared system are concerned. It is worth mentioning here that unlike
to the other controllers. It is also revealed that the MOPID, the previous research in this area, the current investigation was
MOFPI and MOIT2FLC controllers do not have a satisfactory carried out longitudinally to further contribute to the reliabil-
performance in this scenario where there are severe parameter ity of the results concerning the satisfactory functioning of the
changes. proposed controller.

D. Case 4 VII. CONCLUSION

Finally, the last scenario examines the efficiency of the pro- In this paper, a new adaptive and time-varying controller was
posed controller in a situation where there are limitations in presented for the load frequency control of an isolated microgrid.
using EV(s) in the MG. To this end, the EV is disconnected A stochastic multi-objective optimization algorithm was used to
from the grid at t = 47 s. At the same time, the same load dis- optimize the parameter of the controller to track the reference
turbance and wind/PV power fluctuations considered in Cases frequency in the presence of PV, wind generation, V2G elec-
A and B are applied to the system. The simulation results are trical vehicles and load disturbance. To enhance the robustness
presented in Fig. 14. In addition, three common error measure- of the controller against the stated uncertainties of the micro-
ment criteria are used to evaluate the efficacy of the MOPI, grid, it was designed in two levels, namely a Fractional-Order
MOFPI, MOIT2FLC and the suggested control scheme in this Fuzzy and a conventional PI controller. Since the performance
scenario. These included Sum of the Squared Errors (SSE), of the fuzzy systems depends on their membership functions,
Mean Absolute Error (MAE), and Mean Square Error (MSE). the membership function parameters were optimized by using
These approaches can lead to an optimal performance if the val- modification/improvement of MOBHA optimization algorithm.
ues of SSE, MAE, and MSE are close to zero. Table IV presents This novel approach enhances the performance of the LFC with
the evaluation results obtained for these controllers. low computation burden and complexity. In this study, both
The frequency deviations presented in Fig. 14 demon- load disturbance and the output power of DG units were con-
strate that the MOFOFPID controller can better improve the sidered as ΔP in the model of microgrid; hence, the proposed
performance of the MG in comparison to the other three control control approach was found to be adaptive enough to give an
Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 860 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 9, NO. 2, APRIL 2018

appropriate performance in face of uncertain loads, renewable [16] J. Pahasa and I. Ngamroo, “Coordinated control of wind turbine blade pitch
energy sources and MGs of any typology. In order to validate angle and PHEVs using MPCs for load frequency control of microgrid,”
IEEE Syst. J., vol. 10, no. 1, pp. 97–105, Mar. 2016.
the performance and robustness of the controller, Hardware-in- [17] M. H. Khooban, T. Niknam, F. Blaabjerg, and T. Dragičević, “A new
the-Loop (HiL) simulation was used in this study. Furthermore, load frequency control strategy for micro-grids with considering electrical
the performance of the controller was compared with MOPID, vehicles,” Elect. Power Syst. Res., vol. 143, pp. 585–598, Feb. 28, 2017.
[18] V. P. Singh, S. R. Mohanty, N. Kishor, and P. K. Ray, “Robust H-infinity
MOFPI and MOIT2FLC controllers which have been suggested load frequency control in hybrid distributed generation system,” Int. J.
in the most recent research studies on LFC. Elect. Power Energy Syst., vol. 46, pp. 294–305, 2013.
[19] D.-J. Lee and L. Wang, “Small-signal stability analysis of an autonomous
hybrid renewable energy power generation/energy storage system part I:
Time-domain simulations,” IEEE Trans. Energy Convers., vol. 23, no. 1,
ACKNOWLEDGMENT pp. 311–320, Mar. 2008.
[20] E. Mojica-Nava, C. A. Macana, and N. Quijano, “Dynamic population
The authors would like to thank the Iran National Science games for optimal dispatch on hierarchical microgrid control,” IEEE
Foundation for supporting helps to do this work and providing Trans. Syst., Man, Cybern., Syst., vol. 44, no. 3, pp. 306–317, Mar. 2014.
this opportunity for the survey team. Moreover, the authors [21] C. Simon and J. Robert, “Geometric type-1 and type-2 fuzzy logic sys-
tems,” IEEE Trans. Fuzzy Syst., vol. 15, no. 1, pp. 3–15, Feb. 2007.
would also like to thank the reviewers for their very useful [22] M. R. Khalghani, M. H. Khooban, E. Mahboubi-Moghaddam, N.
comments, which helped us to improve the quality of the paper. Vafamand, and M. Goodarzi, “A self-tuning load frequency control strat-
The reviewers’ efforts are gratefully appreciated. egy for microgrids: Human brain emotional learning,” Int. J. Elect. Power
Energy Syst., vol. 75, pp. 311–319, 2016.
[23] M. H. Khooban, T. Niknam, F. Blaabjerg, P. Davari, and T. Dragicevic,
“A robust adaptive load frequency control for micro-grids,” ISA Trans.,
REFERENCES vol. 65, pp. 220–229, Nov. 30, 2016.
[24] H. Bevrani, M. R. Feizi, and S. Ataee, “Robust frequency control in an
[1] B. K. Bose, “Global warming: Energy, environmental pollution, and the
islanded microgrid: HŒ and µ-synthesis approaches,” IEEE Trans. Smart
impact of power electronics,” IEEE Ind. Electron. Mag., vol. 4, no. 1,
Grid, vol. 7, no. 2, pp. 706–717, Mar. 2016.
pp. 6–17, Mar. 2010.
[25] I. Pan and S. Das, Intelligent Fractional Order Systems and Control. An
[2] M. Dehghani, M. H. Khooban, T. Niknam, and S. M. Rafiei, “Time-varying
introduction (Studies in Computational Intelligence), vol. 438. Berlin,
sliding mode control strategy for multibus low-voltage microgrids with
Germany: Springer, 2012.
parallel connected renewable power sources in islanding mode,” J. Energy
[26] S. Das and I. Pan, “On the mixed loop-shaping tradeoffs in fractional-
Eng., vol. 142, no. 4, Feb. 29, 2016, Art. no. 05016002.
order control of the AVR system,” IEEE Trans. Ind. Inf., vol. 10, no. 4,
[3] T. C. Green and M. Prodanović, “Control of inverter-based micro-grids,”
pp. 1982–1991, Nov. 2014.
Elect. Power Syst. Res., vol. 77, pp. 1204–1213, 2007.
[27] J. Zhong and L. Li, “Tuning fractional-order controllers for a solid-core
[4] S. Chowdhury and P. Crossley, Microgrids and Active Distribution Net-
magnetic bearing system,” IEEE Trans. Control Syst. Technol., vol. 23,
works. Stevenage, U.K.: IET, 2009.
no. 4, pp. 1648–1656, Jul. 2015.
[5] J. M. Guerrero, M. Chandorkar, T. Lee and P. C. Loh, “Advanced control
[28] I. Pan and S. Das, “Enhancement of fuzzy PID controller with fractional
architectures for intelligent microgrids part I: Decentralized and hierar-
calculus,” in Intelligent Fractional Order Systems and Control. Berlin,
chical control,” IEEE Trans. Ind. Electron., vol. 60, no. 4, pp. 1254–1262,
Heidelberg: Springer, 2013 pp. 159–193.
Apr. 2013.
[29] M. H. Khooban and T. Niknam, “A new intelligent online fuzzy tuning
[6] H. Jiayi, J. Chuanwen, and X. Rong, “A review on distributed energy
approach for multi-area load frequency control: Self adaptive modified bat
resources and microgrid,” Renewable Sustain. Energy Rev., vol. 12, no. 9,
algorithm,” Int. J. Elect. Power Energy Syst., vol. 71, pp. 254–261, Oct.
pp. 2472–2483, Dec. 2008.
31, 2015.
[7] A. Hooshmand, B. Asghari, and R. K. Sharma, “Experimental demon-
[30] A. Hatamlou, “Black hole: A new heuristic optimization approach for data
stration of a tiered power management system for economic operation of
clustering,” Inf. Sci., vol. 222, pp. 175–184, Feb. 10, 2013.
grid-tied microgrids,” IEEE Trans. Sustain. Energy, vol. 5, no. 4, pp. 1319–
[31] R. Azizipanah-Abarghooee, V. Terzija, F. Golestaneh, and A. Roosta,
1327, Oct. 2014.
“Multiobjective dynamic optimal power flow considering fuzzy-based
[8] R. H. Lasseter and P. Paigi, “Microgrid: A conceptual solution,” in Proc.
smart utilization of mobile electric vehicles,” IEEE Trans. Ind. Informat.,
IEEE 35th Annu. Power Electron. Spec. Conf., 2004, vol. 6, pp. 4285–
vol. 12, no. 2, pp. 503–514, Apr. 2016.
4290.
[32] M. H. Khooban, M. Sha-Sadeghi, T. Niknam, and F. Blaabjerg, “Anal-
[9] J. Yang, Z. Zeng, Y. Tang, J. Yan, H. He, and Y. Wu, “Load frequency
ysis, control and design of speed control of electric vehicles de-
control in isolated micro-grids with electrical vehicles based on multivari-
layed model: Multi-objective fuzzy fractional-order PIλDμ controller,”
able generalized predictive theory,” Energies, vol. 8, no. 3, pp. 2145–2164,
IET Sci., Meas. Technol., vol. 11, pp. 249–261, Nov. 18, 2016. doi:
2015.
10.1049/iet-smt.2016.0277.
[10] M. H. Khooban, T. Niknam, F. Blaabjerg, and M. Dehghani, “Free chatter-
[33] H. Zhang, Y. Zhang, and C. Yin, “Hardware-in-the-loop simulation of
ing hybrid sliding mode control for a class of non-linear systems: Electric
robust mode transition control for a series–parallel hybrid electric vehicle,”
vehicles as a case study,” IET Sci., Meas. Technol., vol. 10, no. 7, pp. 776–
IEEE Trans. Veh. Technol., vol. 65, no. 3, pp. 1059–1069, Mar. 2016.
785, Oct. 1, 2016.
[34] M. Matar, D. Paradis, and R. Iravani, “Real-time simulation of modu-
[11] J. Yang, Z. Zeng, Y. Tang, J. Yan, H. He, and Y. Wu, “Load frequency
lar multilevel converters for controller hardware-in-the-loop testing,” IET
control in isolated micro-grids with electrical vehicles based on multivari-
Power Electron., vol. 9, no. 1, pp. 42–50, Jan. 20, 2016.
able generalized predictive theory,” Energies, vol. 8, no. 3, pp. 2145–2164,
[35] [Online]. Available: www.winddata.com. Accessed on: Oct. 10, 2014.
2015.
[36] [Online]. Available: www.solargis.info/doc/solar-and-pv-data Accessed
[12] S. Zheng, X. Tang, B. Song, S. Lu, and B. Ye, “Stable adaptive PI control
on: Oct. 10, 2014.
for permanent magnet synchronous motor drive based on improved JITL
technique,” ISA Trans., vol. 52, no. 4, pp. 539–549, 2013.
[13] H. Bevrani, F. Habibi, P. Babahajyani, M. Watanabe, and Y. Mitani, “In- Mohammad-Hassan Khooban (M’13) was born in
telligent frequency control in an AC microgrid: Online PSO-based fuzzy Shiraz, Iran, in 1988. He received the Ph.D. de-
tuning approach,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 1935–1944, gree from Shiraz University of Technology, Shiraz,
Dec. 2012. in 2017. From 2016 to 2017, he was a Research
[14] E. Yesil, “Interval type-2 fuzzy PID load frequency controller using big Assistant with the University of Aalborg, Aalborg,
bang–big crunch optimization,” Appl. Soft Comput., vol. 15, pp. 100–112, Denmark conducting research on microgrids and ma-
2014. rine systems. His research interests include control
[15] S. Zheng, X. Tang, B. Song, S. Lu, and B. Ye, “Stable adaptive PI control theory and application, power electronics and its ap-
for permanent magnet synchronous motor drive based on improved JITL plications in power systems, industrial electronics,
technique,” ISA Trans., vol. 52, no. 4, pp. 539–549, 2013. and renewable energy systems.

Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.
 KHOOBAN et al.: LFC IN MICROGRIDS BASED ON A STOCHASTIC NONINTEGER CONTROLLER 861

Taher Niknam (M’14) received the B.Sc. degree Tomislav Dragicevic (S’09–M’13–SM’17) received
from Shiraz University, Shiraz, Iran, in 1998, and the M.E.E. and the industrial Ph.D. degrees from the
the M.Sc. and Ph.D. degrees from Sharif University Faculty of Electrical Engineering, Zagreb, Croatia,
of Technology, Tehran, Iran, in 2000 and 2005, re- in 2009 and 2013, respectively. From 2013 until
spectively, all in power electrical engineering. He is a 2016, he was a Postdoctoral Research Associate at
Faculty Member in the Department of Electrical En- Aalborg University, Denmark. Since March 2016, he
gineering, Shiraz University of Technology. His re- has been an Associate Professor at Aalborg Univer-
search interests include power system restructuring, sity. His field of interest is overall system design of
impacts of distributed generations on power systems, autonomous and grid-connected dc and ac micro-
optimization methods, and evolutionary algorithms. grids, and industrial application of advanced mod-
eling, control and protection concepts to shipboard
power systems, remote telecom stations, domestic and commercial facilities,
and electric vehicle charging stations.

Frede Blaabjerg (S’86–M’88–SM’97–F’03) re-

ceived the Ph.D. degree from Aalborg University,
Aalborg, Denmark, in 1992. From 1987 to 1988, he
was with ABB-Scandia, Randers, Denmark. . He be-
came an Assistant Professor in 1992, an Associate
Professor in 1996, and a Full Professor of power elec-
tronics and drives in 1998. His current research in-
terests include power electronics and its applications,
such as in wind turbines, PV systems, reliability, har-
Mokhtar Shasadeghi received the B.Sc. degree monics, and adjustable speed drives. He received 17
in electronics engineering from Shiraz University, IEEE Prize Paper Awards, the IEEE PELS Distin-
Shiraz, Iran, in 1996, and the M.Sc. and Ph.D. de- guished Service Award in 2009, the EPE-PEMC Council Award in 2010, the
grees from Tarbiat Modares University, Tehran, Iran, IEEE William E. Newell Power Electronics Award 2014, and the VillumKann
in 2001 and 2007, respectively. He is currently an As- Rasmussen Research Award 2014. He was an Editor-in-Chief of the IEEE
sociate Professor at Shiraz University of Technology, TRANSACTIONS ON POWER ELECTRONICS from 2006 to 2012. He has been Dis-
Shiraz. His research interests include robust control, tinguished Lecturer for the IEEE Power Electronics Society from 2005 to 2007
fuzzy control, time-delay systems, linear matrix in- and for the IEEE Industry Applications Society from 2010 to 2011. He is nom-
equalities, and sum- of-squares decomposition and inated in 2014 and 2015 by Thomson Reuters to be between the most 250 cited
optimization. researchers in engineering in the world.

Authorized licensed use limited to: Netaji Subhas University of Technology New Delhi. Downloaded on November 07,2023 at 10:35:31 UTC from IEEE Xplore. Restrictions apply.