Fine-tuning Genetic Algorithm for Photovoltaic-Proton Exchange

Membrane Fuel Cell Hybrid System Optimization

Mustapha Hatti and Hachemi Rahmani

Unité de Développement des Equipements Solaires, UDES, Centre de Développement des Energies Renouvelables, CDER,

Bou Ismail, 42415, W. Tipaza, Algeria

Keywords: PV-PEMFC Hybrid System, Fuzzy Logic Controller, Genetic Algorithm, Management, Hydrogen.

Abstract: European cities have established programs integrating the energy, transport and ICT sectors in order to

deliver more efficient services for their populations. The paper tackles the study of feasibility to implement

fuzzy logic control into an energetic hybrid system and to optimize the membership functions of the fuzzy

logic controller for the Photovoltaic-Proton Exchange Membrane Fuel Cell hybrid system using genetic

algorithm (GA). The paper deals with a fuzzy logic control strategy objective to produce electrical energy

according to the demand, prone to the constraints and the dynamics of the physical load and intermittence of

the energetic resource, by distributing the energy demand between the photovoltaic field and the Proton

Exchange Membrane Fuel Cell system. Photovoltaic-Proton Exchange Membrane Fuel Cell is described in

detail as well as system configuration and components' parameters. The second section devotes to

demonstrating the design process of fuzzy logic control for Photovoltaic-Proton Exchange Membrane Fuel

Cell hybrid System. Finally, the optimal control problem is addressed and genetic algorithm is introduced to

help find a set of optimum parameters in the fuzzy logic controller, best results are obtained and good

optimization of the hybrid system is highlighted.

1 INTRODUCTION

Smart microgrids represent currently an attractive

and viable option for campus applications such as

healthcare, universities, industrial and commercial

complexes, small businesses, residential

neighborhoods and military bases, etc (Dong, 2009).

To avoid problems caused by the weather and

environmental uncertainties, the reliability of a

continuous production of energy from renewable

sources when only one source production system

model is considered, the possibility of integrating

various sources creating hybrid energy solutions can

greatly reduce the intermittences and uncertainties of

energy production bringing a new perspective for the

near future in application on sustainable and smart

cities (Angeliki Kylili, 2015). In the literature

review, a sustainable energy system has been

commonly defined in terms of its energy efficiency,

its reliability, and its environmental impacts

(Alanne, 2006). The basic requirements for an

efficient energy system are its ability to generate

enough power for the world needs at an affordable

price, clean supply, safe and reliable conditions. On

the other hand, the typical characteristics of a

sustainable energy system can be derived from

policy definitions and objectives since they are quite

similar in industrialized countries (Mustapha Hatti,

2011). A hybrid PV-PEMFC low power system is a

suitable solution to replace batteries and to supply

small electric devices placed in remote areas in

particularly in the industrial operations.

The improvement of the efficiency in the energy

production and the guaranty of reliable energy

supply seem nowadays to be common interests of

developed and developing countries (Meriem Naimi-

Ait-Aoudia, 2014). The application of an

autonomous hybrid energy system, typically a

photovoltaic PV-PEMFC hybrid power system, is a

promising solution to electrifying the isolated

locations far from the grid. One of the main

difficulties related to the hybrid structure is the

management of energy flows, (H. Ufuk Gökçe,

2014). Resolution is indeed subject to various

constraints, (Ahmad Atieh, 2015), (Carlos Discoli,

2014), (Punnaiah Veeraboina, 2011), (Forrest

Meggers, 2012), (Padmavathi, 2011), (Qui, 2012).

In recent years, intelligent algorithms have

234

Hatti M. and Rahmani H..

Fine-tuning Genetic Algorithm for Photovoltaic-Proton Exchange Membrane Fuel Cell Hybrid System Optimization.

DOI: 10.5220/0005451702340240

In Proceedings of the 4th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS-2015), pages 234-240

ISBN: 978-989-758-105-2

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

become a popular optimization tool for global and

numerical optimization problems, (Cheng-Hung

Chen, 2008) present a functional-link-based

neurofuzzy network (FLNFN) structure for

nonlinear system control, which is a nonlinear

combination of input variables, using an online

learning algorithm, which consists of structure

learning and parameter learning. Since, the structure

learning depends on the entropy measure to

determine the number of fuzzy rules. The parameter

learning, based on the gradient descent method, can

adjust the shape of the membership function and the

corresponding weights of the FLNN; consequently

they demonstrate the effectiveness of the FLNFN

model, (Vangelis Marinakis, 2013), (Pervez Hameed

Shaikh, 2014), (Marta Maria Sesana, 2015).

2 PV-PEMFC HYBRID SYSTEM

The basic PV-PEMFC structure consists of a

photovoltaic generator, a PEMFC fuel cell and

electrolyzer. With the hybrid structure, three modes

are then possible: Load mode; Normal mode and

overload mode. Charging is the discharge of the fuel

cell takes place independently. (Ming-Feng Han,

2013) propose a group-based differential evolution

algorithm which provides a new process using two

mutation strategies to effectively enhance the search

for the globally optimal solution. Were, all

individuals in the population are partitioned into an

elite group and an inferior group based on their

fitness value. In the elite group, individuals with a

better fitness value employ the local mutation

operation to search for better solutions near the

current best individual. The inferior group, which is

composed of individuals with worse fitness values,

uses a global mutation operation to search for

potential solutions and to increase the diversity of

the population, and GDE algorithm employs

crossover and selection operations to produce

offspring for the next generation.

2.1 Photovoltaïc Subsystem

The electric field created by the p-n junction causes

the photon-generated electron-hole pairs to separate.

The electrons are accelerated to n-region (N-type

material), and the holes are dragged into p-region (P

type material). The electrons from n-region flow

through the external circuit and provide the electrical

power to the load at the same time (Parra, D., 2014).

In this paper a simplified one diode model is

used due its moderate complexity. The relationship

between the output voltage V and the load current I

can be expressed as:

















 1)exp(

IRU

IIIII

oLDL

(1)

Where; I

= light current (A); I

= saturation current

(A); I = load current (A); U = output voltage (V); Rs

= series resistance (Ohm); a = thermal voltage

timing completion factor (V). This model is called

the four parameters model (1); it is simple but

requires determining four parameters value which

are function of temperature, load current and solar

irradiance. The solar panel converts those photons

into electrons of direct current ("DC") electricity.

The electrons flow out of the solar panel and into an

inverter and other electrical safety devices. The

inverter converts that "DC" power (commonly used

in batteries) into alternating current or "AC" power.

AC power is the kind of electrical that electric

device use when plugged into the wall outlet. PV

systems likewise can be blended into virtually every

conceivable structure.

Based on the mathematical equations discussed

before, a dynamic model for a PV module consisting

48 cells in series has been evaluated using

MATLAB/Simulink. The PV array characteristic

presents three important points, the short circuit

current, the open circuit voltage and the optimum

power delivered by the PV to an optimum load when

the PV modules operate at their MPP. The output of

the Matlab function of photovoltaic model

characteristics I/V and P/V is shown first for

different irradiation levels (800; 600; 400; 200

W/m²) at 25°C in figures 2 and 3, and then for

various temperatures (20; 30;40;50 °C) for 800

W/m² in figures 1 and 2 respectively. Results show

excellent correspondence to the model.

Figure 1: I/V characteristics of PV model at different

irradiations and T= 25°C.

0 1020304050

0,0

0,5

1,0

1,5

2,0

2,5

800 (W/m²) @ 25°C

600 (W/m²) @ 25°C

400 (W/m²) @ 25°C

200 (W/m²) @ 25°C

Module Photovoltaic Current (A)

Out

ut Volta

(

)

Fine-tuningGeneticAlgorithmforPhotovoltaic-ProtonExchangeMembraneFuelCellHybridSystemOptimization

235

Figure 2: P/V characteristics of PV model at different

irradiations and T=25°C.

2.2 Fuel Cell Subsystem

The fuel cell, as a renewable energy source, is

considered one of the most promising sources of

electric power. Fuel cells are not only characterized

by higher efficiency than conventional power plants,

but they are also environmentally clean, have

extremely low emission of oxides of nitrogen and

sulfur, and have very low noise.

Fuel cells are electrochemical devices that

convert the chemical energy of a reaction directly

into electrical energy. They have a potential to

achieve a level of efficiency beyond 70% when used

in a cogeneration facility. Fuel cells are classified by

the type of electrolyte used in the cells and include:

(1) proton exchange membrane (polymer) electrolyte

fuel cell (PEMFC), (2) alkaline fuel cell (AFC), (3)

phosphoric acid fuel cell (PAFC), (4) molten

carbonate fuel cell (MCFC), and (5) solid oxide fuel

cell (SOFC). These fuel cells are listed in the order

of approximate operating temperature, ranging from

80°C for PEMFC to1000°C for SOFC.

The typical structure of a single PEMFC is

shown in Figure 3. A single cell consists of anode,

cathode, electrolyte plate and current collectors with

gas channels. H2 and O2 get through the gas

channels of currents collectors and arrive at the

anode and cathode respectively; the reactive gases

pass the diffusion layer and reach the proton

exchange membrane(50 to 170 ptm thick) Figure 3,

membrane under the action of electricity. On the

cathode, the oxygen diffuses towards the catalyst

interface where it combines with the hydrogen

protons and the electrons to form water.

The electrons passing from anode to cathode

produce electrical energy.

Figure 3: Schematic of PEM Fuel Cell.

In the past decades, the researches on PEMFC

are mainly focused on the structural design of a

single cell, catalyst layer and gas diffusion layer, the

manufacture of high performance membrane and

catalyst, the thermal and water management of PEM

fuel cells. The researchers investigated deeply the

components and PEMFC system. Different static

and dynamic models of PEMFC have been

established on the basis of the energy, mass and

momentum conservation laws.

The components and single cell model were

founded based on the operational mechanism, and

research was carried out on the working parameters

(gas flow rate, pressure, humidity, cell temperature

and moisture content) affecting the output voltage.

But the large number of experimental parameters in

the models of components and single cell lead to

overall decrease in performance, figures 4 and 5.

Moreover, the theoretical and simplified

conditions in modeling cause the precision to decline

greatly; and the expressions of model are so

complex that it is difficult to apply them in the

design of PEMFC system.

Figure 4: PEM Fuel Cell Model / MATLAB/Simulink.

0 1020304050

100

110

120

130

140

150

800 (W/m²) @ 25°C

600 (W/m²) @ 25°C

400 (W/m²) @ 25°C

200 (W/m²) @ 25°C

Module Photovoltaic Output Power (W)

Out

ut Volta

(

)

H_ré act io n

Perte de chaleur

H_électricité

Température de sortie

Tension de sortie Vpemfc

3.2e-2

surface de cellule

37.5

0.0

dégradation

Panode

Pcathode

Tension de sortie (V)

Product3

Product2

Product1

Integrator

T ambiante (°

qsl

Chal eur

25-80 °C

-K-

1/2F

-K-

1/(MfcCfc)

0-58V

0-25A

Nombre de cells

Tambiante (°C)

Tinit (°C)

Pcathode (atm)

Panode (atm)

Icharge (A)

SMARTGREENS2015-4thInternationalConferenceonSmartCitiesandGreenICTSystems

236

Figure 5: P/I and V/I Characteristics of PEM Fuel Cell.

2.3 Electrolyser and Hydrogen Tank

Subsystem

An electrolyzer is advice that produces hydrogen

and oxygen from water. The electrochemical

reaction of water electrolysis is given by:

O (l) + elect ene = H

(g) + ½ O

(g) (2)

There are three principal types of water

electrolyser:- alkaline (referring to the nature of its

liquid electrolyte), proton-exchange membrane

(referring to its solid polymeric electrolyte), and

solid-oxide (referring to its solid ceramic

electrolyte), Eq. (2). The alkaline and PEM

electrolysers are well proven devices with thousands

of units in operation, while the solid-oxide

electrolyser is as yet unproven.

The PEM electrolyser is particularly well suited

to highly distributed applications. The alkaline

electrolyser currently dominates global production

of electrolytic hydrogen.

Alkaline water elecrolysis is the dominating

technology today. According to faraday' law,

hydrogen production rate of an electrolyser can be

obtained as:

eCF





(3)

Where: F: Faraday constant, i

: Electrolyser current,

: The number of electrolyser cells in series, η

Faraday efficiency.

: Produced hydrogen moles per second, Eq. (3).

One of the hydrogen storage techniques is physical

hydogen storage, which involves using tanks to store

either compressed hydrogen gas or liquid. The

hydrogen storage model based on Eq. (4) directly

calculates the tank pressure using the ration of

hydrogen flow in the tank. The produced hydrogen

is stored in the tank, whose system dynamic can be

compressed as follow:

bib

RTN

zPP



(4)

Where: M

: Molar mass of hydrogen; N

Hydrogen moles per second delivered to the storage

tank; P

: Pressure of tank.

: Initial pressure of the storage tank; R: Universal

gas constant; T

: Operating temperature; V

: Volume

of the tank; z: Compressibility factor as a function of

pressure, Eq. (4); The hydrogen' state-of-storage

(SHS) is therefore:

maxb

SHS 

(5)

Where: P

: Pressure of tank; P

bmax

: is the maximum

Pressure of the tank, Eq. (5).

3 FUZZY LOGIC CONTROLLER

AND GENETIC ALGORITHM

Since, the wanted behaviour is well known and can

be described using linguistic variables; the use of a

FLC seems appropriate (Shinq-Jen Wu, 2002).

However, it is shown that the method of the

average maximum ensures better performance of

transition (Cheng-Hung Chen, 2008).

It emerges through these comparative studies that

the choice of a better method of defuzzification is

highly dependent on the particular application, and

is the case of the maximum like method which

proves to be very effective for control problems (T.

Azib, 2010).

Simulink block that focused energy management

of a PV-PEMFC hybrid system is implemented:

Figure 6: Energy management system bloc diagram.

This system acts as a fuzzy controller that

controls the operation mode of our hybrid system

regardless of load variations, and photovoltaic

generator, in figure 6. The result is obtained varies

0 5 10 15 20 25

Current (A)

PEM Fuel cell Votlage output (V)

100

200

300

400

PEM Fuel cell power output (Watt)

Fine-tuningGeneticAlgorithmforPhotovoltaic-ProtonExchangeMembraneFuelCellHybridSystemOptimization

237

with the error representing the difference between

load power and power of photovoltaic generator can

therefore defined three operating mode in this

system, namely:

 Normal Mode: In this mode, the power of

positive charge is below the maximum power

from the main source, in this mode the

photovoltaic generator can only feed the load.

 Overload: In this mode the power absorbed by

the load is above the main power source, the

controller can recognize the power of the fuel

cell to the load.

 In Charge Mode (recovery): In this mode, the

power of a PV generator exceeds the load power.

The changes in operating mode only occur when

the load demand is at the boundary of mode

Change. This type of controllers presents many

advantages for this system.

3.1 Genetic Algorithm Optimisation

For the current study, the objective function is the

net power, which is defined as the ratio of the net

electrical power output of the system compared to

the energy power demand to the system, which is

quantified by the value of the variable that is to be

minimized. Summarily, the methodology of the

genetic algorithm consists of the following steps:

1. Generate a random population and evaluate the

fitness of each member;

2. Define the termination conditions;

3. Select parents if the crossover condition is met,

select crossover parameters and apply crossover;

4. Select member if the mutation condition is met,

select mutation parameters and apply mutation;

5. Evaluate the fitness of the offspring and update

the population;

6. Repeat steps 2-6, until the termination conditions

are satisfied.

4 RESULTS AND COMMENTS

Figure 7 shows the residential power demand profile

and solar resources of 48 hours, (Wood Christopher

J. 2010).

The optimum configuration of the hybrid system

proposed is presented and outputs of the Fuzzy logic

controller are depicted in figure.8, following the

behavior of the energetic system. The objective

function has been significantly improved in figures

9-11.

Figure 7: Output of PV generator (PgPV) and load profile

(Pch).

Figure 8: Fuzzy logic control operations.

Figure 9: Load mode and Hydrogen produced.

Figure 10: Overload mode.

0 5 10 15 20 25 30 35 40 45 50

Powers

(

10*e3 x Watts

)

Time

(

Hours

)

P_load (demand)

P_pvg

0 5 10 15 20 25 30 35 40 45 50

0,0

0,2

0,4

0,6

0,8

1,0

Inload mode

Normal mode

Overload mode

Membership degree

Time

(

Hours

)

0 5 10 15 20 25 30 35 40 45 50

Powe

r (1000*Watts), H

(1,5x10*e6 Atm)

Time

(

Hours

)

produced

P_elz

0 5 10 15 20 25 30 35 40 45 5

Powers (10xe*3 Watts)

Time

(

Hours

)

P_pemfc

P_load

SMARTGREENS2015-4thInternationalConferenceonSmartCitiesandGreenICTSystems

238

Figure 11: PEM Fuel Cell power distribution within 48

hours.

The improvement is more noticeable in the

beginning of the optimization process; this flat

behavior indicates that the overall iterative

optimization scheme has practically converged.

5 CONCLUSIONS

In this paper the set model of the stand alone PV-

PEM Fuel Cell hybrid system is analyzed and then

one fuzzy logic controller is considered.

A genetic algorithm is introduced to fine-tune

parameters of membership functions in the fuzzy

logic controller. The results show that the hybrids

with the fine-tuned fuzzy logic controller would

have a higher fuel economy and better system

efficiency compared with the rule-based controller.

The fuzzy logic controller is then designed to

handle with energy distribution and management. To

achieve improved equivalent fuel consumption,

genetic algorithm is implemented to fine-tune the

membership functions.

The control effects must be compared between

different control strategies, e.g. rule-based control

and fine-tuned fuzzy logic control in next future

work.

The results point out those hybrids energetic

systems with the proposed strategy can improve fuel

economy without sacrificing system performance.

REFERENCES

Dong Leilei, Liu Hao, et Riffat Saffa. Development of

small-scale and micro-scale biomass-fuelled CHP

systems–A literature review. Applied thermal

engineering, 2009, vol. 29, no 11, pp. 2119-2126.

Angeliki Kylili, Paris A. Fokaides European smart cities:

The role of zero energy buildings Sustainable Cities

and Society, 15, 2015, pp. 86-95.

Naimi-Ait-Aoudia, M., & Berezowska-Azzag, E. (2014).

Algiers carrying capacity with respect to per capita

domestic water use. Sustainable Cities and

Society, 13, pp. 1-11.

M Hatti, A Meharrar, M Tioursi, Novel approach of

maximum power point tracking for photovoltaic

module neural network based, Int. Symposium on

Environment Friendly Energies in Electrical

Applications, 1-6.

Alanne, K. and Saari, A. Distributed energy generation

and sustainable development. Renewable &

Sustainable Energy Reviews, vol. 10, issue 6, 2006,

pp.539-558.

H. Ufuk Gökçe, K. Umut Gökçe, Multidimensional energy

monitoring, analysis and optimization system for

energy efficient building operations, 2014, Sustainable

Cities and Society, 10, pp 161-173.

Ahmad Atieh, Samir Al shariff, Case study on the return

on investment (ROI) for using renewable energy to

power-up typical house in Saudi Arabia, Sustainable

Cities and Society xxx (2015) xxx–xxx, article in

press.

Carlos Discoli, Irene Martini, Gustavo San Juan, Dante

Barbero, Luciano Dicroce, Carlos Ferreyro, Jésica

Esparza, Methodology aimed at evaluating urban life

quality levels Sustainable Cities and Society, Volume

10, February 2014, Pages 140-148.

Punnaiah Veeraboina, G. Yesuratnam, L. Syam Sundar,

Estimation of Annual Solar Radiation from measured

temperatures by using Temperature-based (TB)

approach in different cities in India Sustainable Cities

and Society, Vol. 1, Issue 4, 2011, pp. 187-194.

Forrest Meggers, Hansjürg Leibundgut, Sheila Kennedy,

Menghao Qin, Mike Schlaich, Werner Sobek,

Masanori Shukuya, Reduce CO

from buildings with

technology to zero emissions, Sustainable Cities and

Society, Vol. 2, Issue 1, 2012, pp. 29-36.

K. Padmavathi, S. Arul Daniel, Studies on installing solar

water pumps in domestic urban sector Sustainable

Cities and Society, Vol. 1, Issue 3, 2011, Pages 135-

141.

Qiu, Guoquan, Shao, Yingjuan, LI, Jinxing, S.B.

Riffat. Experimental investigation of a biomass-fired

ORC-based micro-CHP for domestic applications.

Fuel, 2012, vol. 96, p. 374-382.

Cheng-Hung Chen; Cheng-Jian Lin; Chin-Teng Lin, A

Functional-Link-Based Neurofuzzy Network for

Nonlinear System Control, IEEE Transactions on

Fuzzy Systems, Vol.16, N. 5, 2008, pp. 1362 – 1378.

Vangelis Marinakis, Charikleia Karakosta, Haris Doukas,

Styliani Androulaki, John Psarras, A building

automation and control tool for remote and real time

monitoring of energy consumption, Sustainable Cities

and Society, Vol. 6, 2013, pp. 11-15.

Pervez Hameed Shaikh, Nursyarizal Bin Mohd Nor,

Perumal Nallagownden, Irraivan Elamvazuthi,

Stochastic optimized intelligent controller for smart

energy efficient buildings, Sustainable Cities and

Society, Vol. 13, 2014, pp. 41-45.

0 5 10 15 20 25 30 35 40 45 5

PEMFC-Powers (10xe*3 Watts)

Time (Hours)

P_pemfc-FLC-GA

P_pemfc

Fine-tuningGeneticAlgorithmforPhotovoltaic-ProtonExchangeMembraneFuelCellHybridSystemOptimization

239

Marta Maria Sesana, Branka Cuca, Giuliana Iannaccone,

Raffaella Brumana, Dominique Caccavelli, Charly

Gay, Geomapping methodology for the GeoCluster

Mapping Tool to assess deployment potential of

technologies for energy efficiency in buildings,

Sustainable Cities and Society, Vol. 17, 2015, pp. 22-

34.

Mustapha Hatti, A. Meharrar, M. Tioursi Power

management strategy in the alternative energy

photovoltaic/PEM Fuel Cell hybrid system

Renewable and Sustainable Energy Reviews, vol. 15,

Issue 9, 2011, pp. 5104-5110.

Parra, D., Walker, G. S., & Gillott, M., 2014. Modeling of

PV generation, battery and hydrogen storage to

investigate the benefits of energy storage for single

dwelling. Sustainable Cities and Society, 10, 1-10.

Shinq-Jen Wu; Chin-Teng Lin, Global optimal fuzzy

tracker design based on local concept approach, IEEE

Transactions on Fuzzy Systems, Vol.10, N° 2, 2002 ,

pp. 128 – 143.

Shun-Chung Wang, Yi-Hwa Liu, "A Modified PI-Like

Fuzzy Logic Controller for Switched Reluctance

Motor Drives ," IEEE Trans. on Industrial

Electronics, vol. 58, no. 5, 2011, pp. 1812 - 1825.

Azib T., O. Bethoux, G. Remy, C. Marchand, E.

Berthelot, "An Innovative Control Strategy of a Single

Converter for Hybrid Fuel Cell/Supercapacitor Power

Source", IEEE Trans. on Industrial Electronics, vol.

57, no. 12, 2010, pp. 4024–4031.

Wood, Christopher J., Liu, Hao, et Riffat Saffa B. An

investigation of the heat pump performance and

ground temperature of a piled foundation heat

exchanger system for a residential building. Energy,

2010, vol. 35, no 12, p. 4932-4940.

SMARTGREENS2015-4thInternationalConferenceonSmartCitiesandGreenICTSystems

240