JID: CAEE

ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

Computers and Electrical Engineering 0 0 0 (2018) 1–20

Contents lists available at ScienceDirect

Computers and Electrical Engineering

journal homepage: www.elsevier.com/locate/compeleceng

Multi-objective optimization and energy management in

renewable based AC/DC microgridR
Indragandhi V. a, Logesh R. b,∗, Subramaniyaswamy V. b, Vijayakumar V. c,
Patrick Siarry d, Lorna Uden e
a
School of Electrical Engineering, Vellore Institute of Technology, Vellore, India
b
School of Computing, SASTRA Deemed University, Thanjavur, India
c
School of Computing Sciences and Engineering, Vellore Institute of Technology, Chennai, India
d
Laboratory LiSSi, University of Paris-Est Creteil Val-de-Marne, Creteil, France
e
School of Computing, Staffordshire University, Stoke on Trent, UK

a r t i c l e i n f o a b s t r a c t

Article history: The problems with the design of hybrid micro-grids are system price and service quality. In
Received 18 January 2017 this paper, we solve these problems by utilizing renewable resources optimally, maintain-
Revised 17 January 2018
ing State of Charge (SOC) in batteries. The proposed system also defines the lowest rate for
Accepted 17 January 2018
power exchanged between the AC/DC micro-grids. Photovoltaic and wind energy are uti-
Available online xxx
lized as key resources in the system. Also, storage banks are coupled to both micro-grids
Keywords: and the fuel cell is the hold-up resource to maximize the consistency of the generation
AC micro-grid system. Supervisory controller ensures the maximum utilization of resources and by main-
DC micro-grid taining SOC to manage the exchange of power between micro-grids. This research focuses
MOPSO on power management in AC/DC micro-grid, and its optimization has been investigated
Fuel cell by Multi-Objective Particle Swarm Optimization (MOPSO) algorithm. The result shows that
Photovoltaic MOPSO yields positive performance and the proposed system is recommended as the best
Wind substitute to improve electric energy utilization in remote areas.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction

The increasing rate of population explosion predicts an extra of 2 billion populations by 2040. This up surge of the
population is found mainly in the regions of Africa, India, and several developing countries. This has a huge consequence
on the requirement on energy demand. Energy demand is predicted to surge by 30% in 2040. Developing countries have
witnessed the rapid demand of energy by the population. The predictable energy, like oil and usual gas, hydroelectricity,
coal and nuclear energy are not adequate to meet the need. Furthermore, the population is increasing turning to alternative
energy sources.
Renewable energy sources and the technology to harvest it have grown rapidly. Driven by the economic development
and associated increasing demand for energy, countries are looking for ways to utilize the natural resources available to
meet the demands. Our study is focused on the generation of Photovoltaic (PV) energy in India, as the main energy source
[1]. Complementary nature of wind and PV resources availability are clearly specified for the year 2016 [1]. It shows there

R
Reviews processed and recommended for publication to the Editor-in-Chief by Guest Editor Dr. J. Yogapriya.
∗
Corresponding author.
E-mail address: logeshr@outlook.com (L. R.).

https://doi.org/10.1016/j.compeleceng.2018.01.023
0045-7906/© 2018 Elsevier Ltd. All rights reserved.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

2 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

is urgent need to increase energy asset to advance the extensive scale of renewable based plants. Consequently, PV/Wind
integration is proposed in this research work.
Traditional power system needs more flexibility, altering the regulatory and financial situation. Energy reserves and eco-
logical collision are giving impulsion for the growth of micro-grids, which are forecasted to play an important part in up-
coming power schemes. The micro-grid is a unique procedure of grid which is incorporated with a mixture of dispersed
production and power-electronics. It is our belief that the use of micro-grids with improved performance could provide the
much needs demand of the energy requirements for the Indian populations.
The paper is organized as follows: The next section briefly review relevant literature followed by the detailed presen-
tation of proposed methodology in the Section 3. In the Section 4 we describe the optimization process using MOPSO for
HMG. Section 5 portrays the experiment evaluations carried out to validate the proposed approach along with detailed
discussions on the obtained results. The paper concludes with suggestions for further research.

2. Review on micro-grids technologies

In recent times, micro-grid technology is adopted to organize the different electricity sources to provide enhanced power
supply solutions to the smart-grid based projects in India. Guerrero et al. [2] proposed advanced control technique archi-
tectures for hybrid micro-grids which include dispersed, scattered and categorized control scheme of grid-connected mode
and islanded mode micro-grids. The problems in droop-based control techniques are eliminated in their research. The other
supplementary controller is established in the micro-grid central control which restores the frequency as well as magnitude
in the micro-grid. Sungwoo and Alexis [3] presented a dynamic designing and control stratagem for a micro-grid mainly for
power generated by wind turbine energy and solar energy. A current-fed multi-input source DC–DC type converter is uti-
lized to combine these renewable based energy sources to the key DC bus system [3]. Kantamneni et al. [4] have made an
in-depth survey on the utilization of multi-agent systems for the control of microgrids. They have presented and discussed
the context of multi-agent systems design architectures as step-by-step frameworks for the micro-grids. In 2015, Derksen et
al. [5] have introduced a novel energy agent for future hybrid energy systems for systematic development. They presented a
design pattern to enable the energy agents for the large-scale simulations in the test beds and onsite real-world devices. In
the same year, Dibangoye et al. [6] have explored the unit commitment and economic dispatch problem by determining the
required power to be generated and the optimal schedules as a response to power demand. They also proposed a novel dis-
tributed approach to address the unit commitment problems considering the uncertainty of privacy-preserving constraints
and demand and supply.
Kołodziej et al. [7] presented a solution for scheduling problem in computational grids through utilizing Genetic Algo-
rithms (GA) for effective minimization energy consumption and makespan. They framed the problem as bi-objective in-
cluding flowtime and make span for optimization. Later they presented two variants of improved proved the efficiency and
performances of GAs in the scheduling problem of computational grids. Madiouni et al. [8] developed a novel method named
canonical RST structure for 2DOF, and they also e implemented it successfully based on MOPSO approach. They conducted
different tests over benchmark functions to validate the presented MOPSO based algorithm. With the obtained results from
the conducted experiments, they have proved the superior capability of MOPSO approach through its Pareto dominance. El
Dor et al. [9] presented the dynamic topology called DC luster through combining the Four-cluster and Fitness to avoid the
premature convergence in the traditional PSO algorithm. Their test experiments showed that DCluster has the best perfor-
mance over the other tested traditional topologies for all the tested problems and premature convergence in PSO has very
less probabilities. Similarly, Shenfield and Fleming [10] developed a multi-objective evolutionary design of robust controllers
for the grid and evaluated it through a computer simulation process.
Eghtedarpour and Farjah [11] presented a two-step altered droop control method for the bi-directional power flow of
the hybrid type AC/DC micro-grid. By computing the AC source micro-grid operating frequency (f) and the DC source micro-
grid operating voltage (Vdc ), the electric power management approaches can produce the power reference. The established
power preserve can support both micro-grids regularly, and it also permits minimum preserve power to each micro-grid.
In India, smart grid technologies are increasing consistently with many on-going projects utilizing new smart technology
ranging from kW to MW covering all type of consumers.
The works mentioned above have their significant focus on the DC or AC micro-grid. Technically, the hybrid AC/DC micro-
grid concept is new, and it couples AC sources with AC loads and DC sources with DC loads [14]. It is to be noted that the
uncertainties in the output power of the resources make changes in the generated power to be higher than or lower than
the actual power demand. Such uncertainty in the generated output power leads to problem of fluctuations in the level of
DC bus voltage and hence the DC bus voltage variations should be considered to be an alarming factor for power imbalance
[15]. This crucial problem can be addressed by using a controller to control the DC bus voltage in the power generation
system. The proposed approach is designed to be robust enough against the issues of generation prediction error and the
uncertainties in the resources output power. Although restricting the power exchange between the micro-grids can improve
the power quality, it cannot lead to the maximum utilization of available wind or solar power. Therefore, the trade-off
between the power quality and efficiency of the power generation system is proposed in this paper. The proposed trade-off
determines the minimum value for the amount of power that should be exchanged between the micro-grids.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 3

Fig. 1. Ardent DC-DC boost converter regulated wind system.

3. Proposed renewable based AC/DC micro-grid

Major benefits of the hybrid system AC/DC source micro-grid are: (1) The abolition of preventable multiple alterationpro-
cedures, which results n transmission loss minimization and (2) the removal of entrenched rectifiers for DC heaps in the
existing AC source grids, which results in easyapparatus and cost decline in power electronic converters [14].
Despite the benefits of above-discussed points, for the hybrid micro-grid to work effectively, a superintendent control
technique is required to achieve the power demand and fragmented amongst various energy resources. Also, a fuzzy logic
based control scheme is needed to regulate the charge and discharge currents in the battery. The Hybrid Micro-Grid (HMG)
structure, control and power management strategy is discussed in the following sections.

3.1. Hybrid micro-grid structure

A renewable based hybrid model AC/DC micro-grid and its energy management is proposed in this paper. It is a novel
idea, which disengages DC resources through DC heaps and AC resources with AC heaps, whereas generated power is
swapped amongst both AC and DC edges utilizing a bidirectional mode converter. Depending upon the power exchange
requirement, an AC–DC or DC–AC conversion is used. Photovoltaic array and wind turbine generator are utilized as key re-
sources in this power production system. Also, storage banks are coupled to both AC and DC source micro-grids, and the fuel
cell is considered as a hold-up resource to maximize the consistency of the generation system. When the power delivery
from the input DC sources is larger compared to the DC demand, the inverter operates and inserts power from DC to AC
source sub-grid system and also to utility grid system [16–17]. When the entire power production is smaller than the total
heap demand at the DC source side, the inverter performs as a converter and inserts power from the AC sub-grid system
and utility grid system to DC side. At the same time, DC source bus voltage stabilization is controlled by PI controller.
Micro-grids are a collection of DC and AC heaps, dispersed renewable energy resources and stowagedevices as a battery
[22]. The growth of a hybrid micro-grid will diminish the method of numerous reverse conversions related with individual
AC grid and DC grid [23]. For the proper function of the micro-grid technique, a power management system is necessary
to control electric power stream in the micro-grid scheme. It is imperative that the electric power stream points the iden-
tification of the power output of the production services to attain the power hassle in the load side. It should be noted
that, considering all limitations on power switch-over between the AC and DC micro-grid schemes can develop electric
power quality issues and it cannot direct to highest consumption of existing wind power or solar power [24,25]. So that, a
transaction among electric power system quality and total efficacy of the generation schemes is presented in this research.
The presented wind generation system contains a wind-mill, permanent magnet synchronous generator (PMSG), a diode
rectifier for AC–DC conversion, and a DC–DC boost converter. Block diagram in Fig. 1 shows that the power produced by
wind system is regulated by an ardent DC–DC boost converter. The photovoltaic system is linked via a DC–DC boost con-
verter mode to the DC bus. In order to compute the state of charge of the storage banks in both AC and DC micro-grids,
current addition method (coulomb counting), is employed. In this technique, the norm of calculating the current which
flows inwards and leaving out of the storage is employed. Specifications of fuel cell, PV and wind turbine are listed in Tables
1–3 respectively.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

4 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Table 1
Fuel cell specifications (Horizon H-10 0 0 PEM Fuel Cell).

Content Specification

Number of cells 48
Rated power 10 0 0 W
Performance 28.8 V @ 35A
Required voltage 12 V @ 8A
Efficiency of system 40% at 28.8 V
Low voltage protection 24 V
Over current protection 42A
Max stack temperature 65 °C
Hydrogen pressure 0.45–0.55 Bar
Reactants Hydrogen and Air

Table 2
PV Specifications (Trina Solar TSM-250-PC/PA05A (250 W)).

Content Specification

Rated power 250 W

Type Poly-crystalline
Voc 37.8 V
Isc 8.90 A
Vmpp 30.5 V
Impp 8.20 A
Number of cells 60
Efficiency 15.3%
STC power per unit of area 152.7 W/m2
Maximum system voltage 10 0 0 V

Table 3
Wind turbine specifications (Elsonic India Group).

Objects Specifications

Rated power 1 kW
Maximum power 1.2 kW
Output voltage 220VAC
Type VAWT
Number of blades 5
Rated wind speed 9–11 m/s
Start-up wind speed 3 m/s
Working wind speed 3–25 m/s
Survival wind speed 50 m/s
Output Voltage 48V
Generator efficiency > 0.96

The DC-link stowage capacitor is used to interconnect the resource side and the heaps to common DC source voltage
level. Because the dissimilarities amongst production and load hassle power is the reason for the capacitor stowage voltage
deviations, or the DC source bus voltage range differences, therefore , a regulator is required to normalize the DC source
bus voltage range to the preferred rate. In this research, the storage is used to balance the probable power discrepancies. A
two-level power control technique is employed to find out the charge or discharge source of the storage banks.

3.2. Control of hybrid micro grid

From the control perspective, this research investigates an optimal value for the total source which needs to be swapped
among the AC and DC micro-grids. When the amount of extra electric power from one of the renewable energy source is
smaller than a significant range, the least rate is fixed as zero and source power transfer amongst the AC and DC micro-
grids is not permitted. In distinction, when the amount of electric energy is more than a significant rate, and in the interim,
power deficiency in the next micro-grid source is higher than a predetermined rate, power transfer is allowed. In addition,
the power positions, produced by the system, are also directed to the native power regulator to guarantee pathway of the
assured powers in all subsystem. Also, it should be ensured that the reservations in the renewable resources enable e the
produced power source to be fewer or higher compared to the electric power hassle in a heap.
In the Proportional Integral (PI) and Proportional Integral Derivative (PID) based controllers, the performance of the con-
troller is considerably dependent on the system models. Moreover, the conventional PI and PID based controllers are ex-
tensively parameters dependent and more complicated. The assumptions used in the mathematical model of these systems
drastically reduce the accuracy and affects the performance of the controller. Hence, the conventional PI and PID based con-

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 5

Table 4
Fuzzy rules.

No. Rule

1 if P is PB and SOC is NB, then ib is PB

2 if P is PB and SOC is NS, then ib is PB
3 if P is PB and SOC is ZO, then ib is ZO
4 if P is PS and SOC is NB, then ib is PS
5 if P is PS and SOC is NS, then ib is PS
6 if P is PS and SOC is ZO, then ib is ZO
7 if P is ZO and SOC is NB, then ib is ZO
8 if P is ZO and SOC is NS, then ib is ZO
9 if P is ZO and SOC is ZO, then ib is ZO
10 if P is NS and SOC is NB, then ib is ZO
11 if P is NS and SOC is NS, then ib is NS
12 if P is NS and SOC is ZO, then ib is NS
13 if P is NB and SOC is NB, then ib is ZO
14 if P is NB and SOC is NS, then ib is NB
15 if P is NB and SOC is ZO, then ib is NB

Fig. 2. Input and output variables.

trollers fail to provide improved performance during the changing environment with different load impact and variations in
the parameters.
A fuzzy controller is designed to retain State of Charge (SOC) of the storage system. SOC of battery units are designed
by SOCk . To attain the anticipatedrange of SOC, the fuzzy control technique is designed to be processed in charging period
or discharging period for the presented hybrid type micro-grid. The fuzzy rules are presented in Table 4. Fig. 2 presents
the input variables and output variables for the fuzzy controller. The projected supervisory controller is dedicated only to
establish concluding constructing procedures. The state-chart table demonstrates the projected managerial control, where
S = {S1… S13, S141, S142, S151, S152} is the couple of total nodes in the state-chart,  = {1, …, 12} is the couple of conceiv-
able input actions shown in Tables 5 and 6.
It is stating that S1… S13 is linked to operating modes 1, 2, 3. … 13, correspondingly. The operating mode-14 is cat-
egorized as S141 and S142, in S 141 supremeaccessible energy is greatercompared to power demand, and in S142 supre-
meaccessible energy is smallercompared topower required. Working mode-15 is likewise categorized as S151 and S151. In
S151supreme power is higher than energy requirement and in S151 supreme power is smaller than load power. The descrip-
tions of each state (Modes) in the set S1–S13 are described in Table 7.
The FLC is used to control the SOC of the battery. The fuzzy based SOC controller is compared with the traditional
proportional integral controller to prove the enhanced performance of the fuzzy logic controller, which maintains battery’s

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

6 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Table 5
State-chart Table-I.

Current state (modes) Event Next state (modes)

S1 2 S12
10 S141
6 S2
S2 8 S5
10 S142
5 S1
2 S12
3 S13
S3 12 S151
1 S12
6 S4
S4 12 S152
1 S12
7 S7
4 S13
S5 7 S10
2 S12
S6 1 S5
7 S9
2 S4
S7 8 S11
1 S12
S8 1 S2
2 S7
8 S9

Table 6
State-chart Table-II.

Current state (Modes) Event Next state (Modes)

S9 1 S10
2 S11
S10 2 S12
5 S5
S11 1 S12
5 S13
S12 3 S3
4 S1
S13 8 S6
1 S2
2 S4
7 S8
S141 9 S1
2 S12
6 S142
S142 5 S141
9 S2
S151 11 S3
1 S12
6 S152
S152 5 S131
11 S4

SOC level to provide long life for the battery. The Fig. 3 presents the comparison between the PI and fuzzy logic controller
in a hybrid system. The comparison presents the reduced settling time and maximum overshoot of FLC over PI controller.

3.3. Power management of hybrid micro-grid

From the power managing point of view, the superintendent control for AC source and DC source micro-grid are designed
to perform. Working approaches of the system are classified under four dissimilar operating manners without power transfer
between the micro-grids, and eleven operating modes are defined with power conversation amongst the AC source and DC
source grids, as given below.
With Power Exchange

1. Wind Power satisfies the AC load demand. DC loads are met by Wind Power and Solar Power.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 7

Table 7
State description with respect to events.

Event Description

1 WG
Pmax ≥ PLAC
2 PV
Pmax ≥ PLDC
3 WG
Pmax ≤ PLAC
4 PV
Pmax ≤ PLDC
5 WG
Pmax + Pmax
PV
≥ PLAC + PLDC
6 WG
Pmax + Pmax
PV
< PLAC + PLDC
7 SOCAC < 0.2
8 SOCDC < 0.2
9 PWG ≥ PCritical
WG

10 PWG < PCritical

WG

11 PPV ≥ PCritical
PV

12 PPV ≥ PCritical
PV

Fig. 3. Comparison of PI and FLC.

2. Wind Power satisfies the AC load demand. DC loads are met by Wind Power and Solar Power along with the battery.
3. Wind Power and Solar Power satisfies AC load demand. DC load demand is met by solar power.
4. Wind Turbine Power and PV Power along with storage system satisfy AC source loads. DC source loads are satisfied by
PV power.
5. Wind power and battery satisfies the AC load demand, solar and battery satisfies the DC load demand.
6. Wind Power and Solar Power along with battery satisfies DC micro grid meets AC load demand. DC loads are satisfied
by solar power.
7. Wind power and battery in DC micro grid satisfies the AC load demand, solar and battery in DC micro grid satisfies the
DC load demand.
8. Wind power and Fuel cell satisfies the AC load demand. Solar system and Fuel cell satisfies the DC load demand
9. Wind power satisfies the AC load demand. Solar system and Fuel cell satisfies the DC load demand
10. Wind power satisfies the AC load demand. Solar system, wind power, and Fuel cell satisfies the DC load demand
11. Solar power satisfies the DC load demand. Solar system, wind power, and Fuel cell satisfies the AC load demand
Without Power Exchange
1. AC load demand met by wind power. DC source heap hassle met by solar energy.
2. AC load demand met by wind power with battery. DC source heap hassle encountered by the solar source with battery.
3. AC load demand met by wind power. DC load hassle met by the solar source with battery.
4. AC load demand met by wind power with battery. DC load hassle met by solar power.
The foremost concern is to reduce consumption of the fuel cell to a maximum and limit handling of the key converter.
Also, the proposed power management system is used to reduce system running cost and fuel expenditure. Although power

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

8 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Fig. 4. Flow Chart based on AC availability.

management systems are supposed to include better performance, production forecast errors are not measured in the mod-
eling process which could have an effect on the efficiency of the system.
The Power Management System (PMS) contains a dominant processing unit with the data obtained for the entire micro-
grid modules, which helps to find effective approaches aimed at power transfer among renewable resources. The smallest
and extreme charge/discharge mode energy of the storage tank, and lowest permissible power interchange among the AC
source and the DC source micro-grids, ought to be informed based on real-time measurements. Likewise, output instructions
of the PMS are reference-points for production units, charge/discharge source power of the storage tank, and the quantity of
power source must be switched within the micro-grids. The delivery instructions are directed to the controllers to enforce
the micro-grid system modules to follow the attained reference-points. The power flow between AC and DC grids using
bi-directional converter is shown in Fig. 4 when wind energy alone is considered. The power flow between AC and DC grids
using bi-directional converter is provided in Fig. 5 when wind energy alone is considered.
Implementation of the above approach produces huge benefits. These include: (i) maximum usage of renewable resources
fulfilling the AC load demand and DC load demand in micro-grid system, (ii) SOC of battery is sustained together in micro-
grids and (iii) Power transfer among the AC source and the DC source micro-grids is supervised.

4. Optimization using MOPSO for HMG

Hybrid Micro-Grid System (HMGS) offers an optimal, reliable, and cost-effective solution for utilizing localized renewable
energy resources over individual DC or AC micro grid. Generally, production, distribution, and demand subsystems are joined
together to form an HMG, that vary greatly depending on availability of renewable resources, desired services to provide, and
demand subsystem parameters. These parameters together have a high impact on decision making, reduction of the cost and
to improve reliability of the system. MOPSO is used in this research to obtain the power management approach for a HMG
System to retain an uninterrupted power to the load demand in different modes of operation. This technique is realistic in
order to get the pre-eminent configuration of power system and for equipment’s sizing [12,13]. The PSO is inspired from the
evolutionary computing field and the social interactions of the swarm and therefore formed the swarm intelligence model
[18,19]. Typically, in the PSO algorithms, each position of every particle, there are two best values determined as individual
best and global best [20,21]. The individual best can be defined as the best value of the particle which is achieved and
stored. The global best is defined as the best value of all particles of the population and it is determined by the PSO
optimizer. The position of each particle represents the values of the variable and their velocity leads them to attain global
and individual bests. The objective function which is used to determine the best solution of all feasible solutions is called
as the fitness function. As the personalization of PSO algorithm, the constraints may be included with the fitness to attain

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 9

Fig. 5. Flow Chart based on DC availability.

better optimization.
 

k
f i (x ) 
k
fitness = min wi max with wi ≥ 0 and wi = 1 (1)
fi
i=1 i=1

Minimize
Cost (APV , AW ind , PCapBatt ,  , AF C ) (2)

Maximize
A (APV , AW ind , PCapBatt ,  , AF C ) (3)

The three major steps of PSO algorithm is as follows:

• Compute and evaluate the fitness of each particle through fitness function.
• Update the global and individual best position and fitness accordingly.
• Update the position and velocity of each particle.

During the operation of the PSO algorithm, the achieved best fitness value for each particle is stored, and the particle
with the best fitness value is compared with the other particle. As the comparison is made, the fitness value is updated
during the iterations of the algorithm’s operation. The iterative process is terminated with the stopping criteria to meet the
values of the predefined target fitness.
For each particle of the swarm population, the position is updated through the following equation.

posind ind ind

par _index+1 = pos par _index + vel par _index+1 (4)
Here, pos is the position of the variable and vel represents the velocity during the par_index.
The velocity vel of the particle is determined using the following equation.
 ind   glo 
par _index+1 = ω .vel par _index + coe f f 1.rand1 par par _index − pos par _index + coe f f 2.rand2 par par _index − pos par _index
vel ind ind ind ind
(5)

Here, parind is the individual best position and parglo is the global best position. coeff1 is the cognitive parameter and
coeff2 is the social parameter. In the above equation, the random numbers ranging between 0 and 1 is represented by rand1
and rand2. The values of coeff1 and coeff2 mostly close to 2 and it affects the size of the particle’s move towards global
and individual bests. For our study, we have assumed the values for both coeff1 and coeff2 to 2 for the equal attraction of

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

10 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Fig. 6. Working framework of MOPSO algorithm.

particles to the best points. ωis known as the inertia weight, which helps in the movement of the particle with the same
velocity and same direction.
coe f f 1.rand1( par ind
par _index
− posind
par _index
)is the cognitive component which influences the particle to revisit the former po-
sition where it has experienced the higher fitness value.
glo
coe f f 2.rand2( par par_index − posind
par _index
)is known as the social component that helps the particle to return to swarm’s
best region which has been found so far. It is to be noted that when is coeff1 >> coeff2, then the particles are attracted to
the individual best positions more and when is coeff2 >> coeff1, then the particles are better attracted to the global best
positions. MOPSO algorithm differs from the traditional PSO algorithm by creating an external archive for managing the
individual and global bests. The utilization of inherent characteristics is to track the particles in the swarm and optimal
selection mechanism of PSO with MOPSO helps to trace the dominated solutions.
The exploitation of PSO algorithm for the multi-objective problems has considerable advantages. Similar to other Evo-
lutionary Algorithms (EA) and Genetic Algorithms (GA), PSO algorithm is based on the sizable population of individuals or
particles. PSO hunts for the non-dominated solution in the solution space of individuals or particles. As the PSO trials after
the best position’s direction, the global optimal particle is in the memory in some sense. In that case, PSO has considerable
efficiency. The working framework of MOSPO is depicted in Fig. 6.
From Fig. 6, it can be noted that MOPSO algorithm differs from the traditional PSO algorithm by creating an external
archive for managing the individual and global bests. The utilization of inherent characteristics is to track the particles in
the swarm and optimal selection mechanism of PSO with MOPSO helps to trace the dominated solutions.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 11

Fig. 7. Change in DC load in one minute.

4.1. MOPSO algorithm

Initialize
Load data (solar radiation and wind speed etc.)
Load the characteristics of the components
Load the economical parameters
For each par
generate pos and vel randomly
update pos and vel as per posindpar _index+1
and vel ind
par _index+1
Evaluate using objective functions (mEC and PLP)
end For
update the External_archive
While (until stopping criterion)
for each par in swarm
obtain parind and parglo
update pos and vel as per posind par _index+1
and vel indpar _index+1
end For
For each par in External_archive
trim the External_archive
end For
update the External_archive
end While
return External_archive

From our example, we have defined our fitness function as Minimum Energy Cost (MEC) and Power Loss Probability
(PLP). In the iteration of the algorithm, mEC and PLP were calculated for the each generated particles and if the constraint
is met then it is accepted as the PSO particles of the population. We have used MATLAB to manage the power and operations
of the presented hybrid micro-grid system.

5. Experimental evaluations and discussions of Energy Management

The developed system with MOPSO algorithm was experimentally evaluated to determine its performance over the multi-
objective fitness functions. Fig. 7 shows the DC load demand variation in one minute. The peak demand was met at the time
period of 30–40 s. In the same time period the power generated from the PV is also high which may be utilized to satisfy
the AC micro-grid load requirement (Fig. 8). The SOC of DC side battery is shown in Fig. 9. It indicates that the discharging
of the battery takes place at the initial time period of 0–11 s. From 11 to 40 s there was no operation of the battery due to
minimum power availability. At last the excess power is utilized for battery charging from 40 to 60 s.
Fig. 10 shows the AC side load variations in one minute. The plot depicts that the peak demand is at the time period
of 28–47 s. The minimum load requirement of 10 0 0 W is met at starting and ending time of the single minute plot. In the
same time period the power generated from the wind is also high which may be utilized to satisfy the DC micro-grid load

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

12 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Fig. 8. Change in PV power in one minute.

Fig. 9. Change in SOC of the DC side battery one minute.

requirement (Fig. 11). The SOC of DC side battery is shown in Fig. 12. It shows that the discharging of the battery takes place
at the initial time period of 0–9 s. From 9 to 42 s there is no operation of the battery due to minimum power availability.
At last the excess power is utilized for battery charging from 42 to 60 s.The power profile from AC to DC; also DC to AC is
presented in Figs. 13 and 14.

5.1. Experiments on PSO controller based DC-DC converter

Unfortunately, the characteristic of renewable sources are having low DC voltage in nature and is not suitable for direct
utilization in the micro-grid. In general, photovoltaic modules are connected in series in order to reach high voltage values.
This in turn requires large physical place and number of PV modules. To overcome that and to convert low voltage to high
voltage, efficient DC–DC converter is required which makes better utilization of renewable energy sources. Similarly fuel cell
and wind energy conversion system also require DC–DC converter. The converter must adhere to the requirements such as

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 13

Fig. 10. Change in AC load one minute.

Fig. 11. Change in wind power one minute.

low cost, low weight, reduced witching voltage stress, and high power density. In this paper a PSO based non-isolated high
step up converter to generate high voltage gain using a single switch is proposed (Fig. 15). The advantages of this converter
are high voltage conversion, low conduction loss, low voltage stress across the semiconductor devices, fewer components
and easy control.
The proposed converter output voltage is fed to the single phase full bridge inverter, which converts 305 Vdc into peak
to peak 305 Vac as shown in Fig. 16. The proposed converter produces the output voltage of 305 Vdc which is suitable for
220 V (RMS) AC voltage grid applications. The output current from the converter is shown in Fig. 17. The proposed converter
is designed with single switch and voltage across this switch is presented in Fig. 18. Input capacitor voltage charging and
discharging performance is provided in Fig. 19. The current passing through the input side diode and input side inductor is
presented in Figs. 20 and 21.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

14 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Fig. 12. Change in SOC of the AC side battery one minute.

Fig. 13. Power from AC to DC.

Compared to the higher performance and simplicity of sliding mode algorithm, PID control algorithm is simple and has
lower adaptability. In addition, the tuning process of the controller is found to be bit lengthy due to the trial and error
method. Even utilization of effective optimization algorithms needs time to define optimal parameters for the controller
based on its specific characteristics. When there are any changes in the system, then the controller needs re-tuning to en-
sure the delivery effective system response. Based on the above-mentioned limitations, the need for the efficient control
algorithm is high and deployment of new algorithm is focused on delivering minimal or no tuning based control algorithm
with maximum adaptation and improved performance in terms of overshoot and settling time. In this work, we explore the
possibilities of utilizing PSO for controller algorithm in a DC–DC converter based on its characteristics and features includ-
ing robustness, self-adaptability, the simplicity of implementation, versatility. Rather exploiting the PSO for optimizing the
parameters of the controller, we intend to use PSO as a controller for the DC–DC converter. The proposed PSO controller
algorithm computes Converter Modulation Index, which is needed to map the output voltage to reference voltage by re-
ducing the error. Fig. 22 portrays the comparison of different controllers with respect to step response specification of the
undistributed system.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 15

Fig. 14. Power from DC to AC.

Fig. 15. Non-isolated high step-up converter.

Fig. 16. DC output voltage.

Fig. 17. Converter DC output current.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

16 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Fig. 18. Voltage across the single switch.

Fig. 19. Input capacitive voltage.

Fig. 20. Input diode current.

Fig. 21. Input inductor current.

In Fig. 23, the movement of the swarms for 120 iterations is presented. The Fig. 23 shows that all the particle swing
from initial unsystematic position to the best solution. Therefore all the particles join to a single position which is known as
universal best. Subsequently, the ideal groupings are situated in certain distant positions from one another with the similar
capability and dissimilar structures in the objective area, planning such schemes are a difficult assignment. However, the
particles come near to each other later than 120 iterations, and the finest mixture is recognized. Furthermore, the outcomes
for particular automatic runs nearly join to the identical ideal fitness mark. For finding a superior judgment, the number of
positions on visible should be improved by difference or supplementary nonlinear functions which are offered in upcoming
works. From the obtained experimental results, it is clear that MOPSO provides best solar, wind and battery ratings. The
optimal solutions of MOPSO are presented in the Table 8. The optimization model of the MOPSO generates suitable sizing
for the opted location. The location that is selected from Tamilnadu, India has temperate solar radiation and high wind
speed. The results obtained from the MOPSO depict the improved reliability with reduced cost with maximized contribution
of renewable sources for the selected location. The utilization of the renewable energy in the selected location enhances the
energy accessibility by the poor communities and improves the standards of living. With the increase of PV panels, the
system’s Loss of Power Supply Probability (LPSP) is maximized and hence the upper limit remains the same.
The proposed system with MOPSO is compared with differential evaluation and genetic algorithm schemes and presented
in Table 9.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 17

Fig. 22. Comparison of different controllers with respect to step response specification of the undistributed system.

Fig. 23. Simulation process of MOPSO algorithm.

Table 8
MOPSO result.

Parameters Value

Number of iterations 120

Number of particles 120
Number of PV panels 60
wind turbines 1.2kW
Fuel cell 250W
LPSP (%) 8
COE ( /kWh) 1.79
Renewable factor (%) 60

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

18 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

Table 9
Comparison of differential evolution, MOPSO and genetic algorithm.

Comparison features Differential evolution Multi-objective particle swarm Genetic algorithm

optimization

Impact of population size over solution time Linear Linear Exponential

Effect of best solution with respect to population Less Most Medium
Average best fitness Best Good Good
Need for the ranking of solutions No No Yes
Search space density More More Less
Possibility of premature convergence Low High Medium
Improvement of convergence with homogeneous NA Yes Yes
sub-grouping
Average best fitness Best Good Good

6. Conclusions and future works

In the rural electrification, the micro-grid implementation is considered to be a most promising solution which increases
the supply quality by reducing the implementation costs. A control scheme for an HMG Structure to sustain uninterrupted
power supply to the consumer demand in fifteen modes of operation is presented in this research. The combination of PV,
Fuel cell, wind and battery storage with adjustable parameters is analysed. By utilizing the state machine approach, the
supervisory controller sets the reference values to the generation subsystems through following a pre-defined scheme. The
fuzzy controller determines the charge and discharge of battery banks by considering the difference between the gener-
ated and demanded power, and SOC. To achieve best configuration the system and better sizing of components, we have
applied MOPSO methodology by defining the objective functions for energy cost and power loss probability. The simula-
tion results depict the price increase of electricity which leads to the substantial growth in utilization of HMG based on
renewable resources. Hence using renewable resources is a worthy substitute to develop the energy power in isolated areas
in India. The presented methodology can be extended to develop a modernized electrical grid utilizing advanced informa-
tion and communication technologies (ICT). The incorporation of ICT to analyse the behaviour of consumers and suppliers
through automation with artificial intelligence may lead to the tremendous benefits of an efficient, reliable, and sustainable
production and distribution of electricity.

Acknowledgement

The authors are grateful to Science and Engineering Research Board (SERB), Department of Science & Technology, New
Delhi, for the financial support (No. YSS/2014/0 0 0718/ES).

References

[1] http://www.nrel.gov/international/images/india_dni_annual_2016-01.jpg.
[2] Guerrero JM, Loh PC, Lee TL, Chandorkar M. Advanced control architectures for intelligent microgrids-part II: power quality, energy storage, and AC/DC
microgrids. IEEE Trans Ind Electron 2013;60(4):1263–70.
[3] Sungwoo B, Alexis K. Dynamic modeling and operation strategy for a microgrid with wind and photovoltaic resources. IEEE Trans Smart Grid 2012;3(4).
[4] Kantamneni A, Brown LE, Parker G, Weaver WW. Survey of multi-agent systems for microgrid control. Eng Appl Artif Intell 2015;45:192–203.
[5] Derksen C, Linnenberg T, Unland R, Fay A. Structure and classification of unified energy agents as a base for the systematic development of future
energy grids. Eng Appl Artif Intell 2015;41:310–24.
[6] Dibangoye J, Doniec A, Fakham H, Colas F, Guillaud X. Distributed economic dispatch of embedded generation in smart grids. Eng Appl Artif Intell
2015;44(2338):64–78 art..
[7] Kołodziej J, Khan SU, Xhafa F. Genetic algorithms for energy-aware scheduling in computational grids. In: IEEE international conference on P2P, parallel,
grid, cloud and internet computing; 2011. p. 17–24.
[8] Madiouni R, Bouallègue S, Haggège J, Siarry P. Robust RST control design based on Multi-Objective Particle Swarm Optimization approach. Int J Control,
Autom Syst 2016;14(6):1607–17.
[9] ElDor A, Lemoine D, Clerc M, Deroussi L, Gourgand M. Dynamic cluster in particle swarm optimization algorithm. Nat Comput 2015;14(4):655–72.
[10] Shenfield A, Fleming PJ. Multi-objective evolutionary design of robust controllers on the grid. Eng Appl Artif Intell 2014;27:17–27.
[11] Eghtedarpour, Farjah E. Power control and management in a hybrid AC/DC microgrid. IEEE Trans Smart Grid 2014;5(3):1494–505.
[12] Kim M, Bae S. Decentralized control of a scalable photovoltaic (PV)-battery hybrid power system. Appl Energy 2017;188:444–55.
[13] Kornelakis A. Multiobjective particle swarm optimization for the optimal design of photovoltaic grid-connected systems. Sol Energy
2010;84(12):2022–33.
[14] Wang P, Geol L, Liu X, Choo FH. Harmonizing AC and DC: A hybrid AC/DC future grid solution. IEEE Power Energy Mag 2013;11(3):76–83 May/Jun..
[15] Wu TF, Kuo CL, Sun KH, Chang YC. DC-bus voltage regulation and power compensation with bi-directional inverter in dcmicro-grid applications. In:
Proc. IEEE Energy Convers. Congr. Expo. (ECCE), Sep. 17–22; 2011. p. 4161–8.
[16] Sechilariu M, Wang BC, Locment F. Supervision control for optimal energy cost management in DC microgrid: design and simulation. Int J Electr Power
2014;58:140–9.
[17] Chen YK, Wu YC, Song CC, Chen YS. ‘Design and implementation of energy management system with fuzzy control for DC micro-grid systems’. IEEE
Trans Power Electron 2013;28(4):1563–70.
[18] Lee Y-L, Chen K-Y, Liao S-J. Using proportional navigation and a particle swarm optimization algorithm to design a dual mode guidance. Comput Electr
Eng 2015 Article in Press.
[19] Lin C-J, Li T-HS, Kuo P-H, Wang Y-H. Integrated particle swarm optimization algorithm based obstacle avoidance control design for home service robot.
Comput Electr Eng 2016;56:748–62.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20 19

[20] Chen K-Y, Lee Y-L, Liao S-J, Kung C-C. The design of particle swarm optimization guidance using a line-of-sight evaluation method. Comput Electr Eng
2016;54:159–69.
[21] Azharuddin M, Jana PK. Particle swarm optimization for maximizing lifetime of wireless sensor networks. Comput Electr Eng 2016;51:26–42.
[22] Adefarati T, Bansal RC. Reliability and economic assessment of a microgrid power system with the integration of renewable energy resources. Appl
Energy 2017;206:911–33.
[23] Ashray M, Giri V, David S. Simple electric utility platform: A hardware/software solution for operating emergent microgrids. Appl Energy
2018;210:748–63.
[24] Tohid S, Daryoush N, Sajjad G. A multi-objective HEM strategy for smart home energy scheduling: A collaborative approach to support microgrid
operation. Sustainable Cities Soc 2018;37(February):26–33.
[25] Maulik A, Das D. Optimal operation of microgrid using four different optimization techniques. Sustainable Energy Technol Assess 2017;21:100–20.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023
JID: CAEE
ARTICLE IN PRESS [m3Gsc;January 24, 2018;8:27]

20 I. V. et al. / Computers and Electrical Engineering 000 (2018) 1–20

V. Indragandhi completed B.E. in Electrical and Electronics Engineering from Bharathidasan University, Tiruchirapalli in 2004 and received M.E. in Power
Electronics and Drives from Anna University, Chennai in 2010. She obtained the Ph.D. degree from Anna University, Chennai in the year 2015. At present
she is an Associate Professor in the School of Electrical Engineering, VIT, Vellore, India.

Logesh R. received the B.Tech. degree in Computer Science & Engineering and M.Tech. degree in Networking from Pondicherry University, India. Currently,
he is a Junior Research Fellow and pursuing the Ph.D. degree in the School of Computing, SASTRA Deemed University, India. His research interests include
Recommender Systems, Information Retrieval, and Human–Computer Interaction.

Subramaniyaswamy V. received the B.E. degree in Computer Science & Engineering and M.Tech. degree in Information Technology from Bharathidasan
University, India and Sathyabama University, India respectively. He received his Ph.D. degree from Anna University, India and currently working as a Senior
Assistant Professor in the School of Computing, SASTRA Deemed University, India. His research interests include Data Mining and Recommender Systems.

Vijayakumar V. is currently working as Professor in VIT, Chennai Campus; He has more than 18 years of experience which includes 8 years in teaching
and 8 years in Industry. He is also a coordinator of Cloud Computing Research Group and Coordinator for Internship in India and Worldwide. His area of
research includes Grid Computing, Cloud computing and Big Data.

Patrick Siarry received the Ph.D. degree from the University Paris 6, in 1986 and the Doctorate of Sciences (Habilitation) from the University Paris 11, in
1994. He was first involved in the development of analog and digital models of nuclear power plants at Electricité de France (E.D.F.). Since 1995, he is a
professor in automatics and informatics.

Lorna Uden is Emeritus Professor of IT systems in the Faculty of Computing, Engineering and Technology at Staffordshire University. She is on the editorial
board of several international journals. She published widely in conferences, journals, books and workshops. She is program committee member for many
international conferences and workshops. She is the founder of the KMO and LTEC conferences.

Please cite this article as: I. V. et al., Multi-objective optimization and energy management in renewable based AC/DC
microgrid, Computers and Electrical Engineering (2018), https://doi.org/10.1016/j.compeleceng.2018.01.023