Modeling and optimization for proton exchange membrane fuel cell
stack using aging and challenging P systems based optimization
algorithm
Shipin Yang
a
,
b
,
*
, Ryad Chellali
a
, Xiaohua Lu
b
, Lijuan Li
a
, Cuimei Bo
a
a
College of Electrical Engineering and Control Science, Nanjing Tech University, Nanjing, 211816, China
b
State Key Laboratory of Materials-Oriented Chemical Engineering, Nanjing Tech University, Nanjing, 210009, China
article info
Article history:
Received 26 November 2015
Received in revised form
11 March 2016
Accepted 21 April 2016
Available online 27 May 2016
Keywords:
PEM fuel cell
Modeling
P systems
Aging
Parameter estimation
abstract
Accurate models of PEM (proton exchange membrane) fuel cells are of great significance for the analysis
and the control for power generation. We present a new semi-empirical model to predict the voltage
outputs of PEM fuel cell stacks. We also introduce a new estimation method, called AC-POA (aging and
challenging P systems based optimization algorithm) allowing deriving the parameters of the semi-
empirical model. In our model, the cathode inlet pressure is selected as an additional factor to modify
the expression of concentration over-voltage V
con
for traditional Amphlett's PEM fuel cell model. In AC-
POA, the aging-mechanism inspired object updating rule is merged in existing P system. We validate
through experiments the effectiveness of AC-POA and the fitting accuracy of our model. Modeling
comparison results show that the predictions of our model are the best in terms of fitting to actual
sample data.
© 2016 Elsevier Ltd. All rights reserved.
1. Introduction
The PEM (proton exchange membrane) fuel cell is a low tem-
perature electrochemical device that offers a promising and
possibly green alternative to traditional power sources and other
fuel cell types in many applications with avoiding air polluting is-
sues [1e3]. By operating on principle of the electrochemical reac-
tion between hydrogen and oxygen, with the aid of a catalyst
(platinum) [4], it can produce current directly.
For a better understanding of the phenomena occurring within
PEM fuel cells, several models have been proposed over the last
decade [5]. Some of these models gave rise to important founda-
tions for PEM fuel cells’ control technologies. Existing modeling
approaches can be divided into two main categories. The first
category deals with mechanistic modeling: it involves heat, mass
transfer, and electrochemical phenomena in the modeling process
[6e8]. The second category relies on empirical or semi-empirical
equations, which are partly inspired by mechanistic models [7].
Among existing PEM fuel cell models, the one proposed by
Amphlett et al. [9] seems widely accepted by the community. In this
model, the output voltage V
cell
of a single cell is described by the
reversible voltage V
Nernst
, activation overpotential V
act
, ohmic
voltage drop V
ohmic
, and the diffusion overpotential V
con
.
According to Amphlett's model, many researchers addressed the
PEM fuel cell modeling as a fitting problem, e.g., the search of the
parameters set minimizing distances between actual and predicted
data. Ohenoja et al. [7] adopted a real-coded genetic algorithm and
series of sample data to estimate the best parameters for PEM fuel
cell. Zhang et al. [10] obtained the optimal model using an adaptive
RNA (ribonucleic acid) genetic algorithm. Mo et al. [11] used a niche
HGA (hybrid genetic algorithm) to determine and optimize the PEM
fuel cell parameters. Askarzadeh et al. proposed series of optimi-
zation algorithms (such as BMO (bird mating optimizer) [12], MPSO
(modified particle swarm optimization) [13], ABSO (artificial bee
swarm optimization algorithm) [14], GGHS (grouping-based global
harmony search algorithm) [15], IGHS (innovative global harmony
search algorithm) [16]) to obtain optimal parameters. Considering
modeling results,it can be seen clearly that all models can accu-
rately describe the iev characteristics of the PEM fuel cell at low
and middle current density segments under various pressures.
However, when it comes to the high current density workspace,
these models deviate and errors occur gradually in all cases.
* Corresponding author. College of Electrical Engineering and Control Science,
Nanjing Tech University, Nanjing, 211816, China.
E-mail address: spyang@njtech.edu.cn (S. Yang).
Contents lists available at ScienceDirect
Energy
journal homepage: www.elsevier.com/locate/energy
http://dx.doi.org/10.1016/j.energy.2016.04.093
0360-5442/© 2016 Elsevier Ltd. All rights reserved.
Energy 109 (2016) 569e577