Thermal and Electrical Parameter Identification of a Proton Exchange Membrane Fuel Cell Using Genetic Algorithm

Abstract

PEM fuel cell is a technology successfully used in the production of energy from hydrogen, allowing the use of hydrogen as an energy vector. It is scalable for stationary and mobile applications. However, the technology demands more research. An important research topic is fault diagnosis and condition monitoring to improve the life and the efficiency and to reduce the operation costs of PEMFC devices. Consequently, there is a need of physical models that let deep analysis. These models must be accurate enough to represent the PEMFC behavior and to allow the identification of different internal signals of a PEM fuel cell. This work presents a PEM fuel cell model that uses the output temperature in a closed loop, so it can represent the thermal and the electrical behavior. The model is used to represent a NEXA Ballard 1.2 kW; therefore it is necessary to fit the coefficients to represent the real behavior. Five optimization algorithms were tested to fit the model, three of them were taken from literature and two were proposed. Finally, the model with the parameters identified was validated with real.

A simplified dynamic model with driver’s NMS characteristic for human-vehicle shared control of autonomous vehicle

“…The HVSC model parameters are identified by AGA method where optimal identified value can be approached through continuous generation of iterations. 26,27…”

Adaptive Online Parameter Estimation Algorithm of PEM Fuel Cells

“…Fuel cells have been considered a promising and environmental-friendly energy source in recent years [1]. Through the chemical reaction between hydrogen and oxygen, fuel cells can directly convert the chemical energy into water, electricity and heat [2]- [5]. Based on this chemical principle, there are various advantages of fuel cells, such as high efficiency, zero pollutant emission, and flexible installation, etc.…”

“…1 Institut de Robòtica i Informàtica Industrial, CSIC-UPC Llorens i Artigas 4-6, 08028 Barcelona, Spain (Email: yashan.xing@upc.com, ramon.costa@upc.edu) 2 Faculty of Mechanical & Electrical Engineering, Kunming University of Science & Technology, Kunming, 650500, P.R. China.…”

“…However, these approaches assume that noise or disturbance is uncorrelated and independent of the regression vector. In [2], [16], genetic algorithm combined with curve fitting procedure was used to estimate fuel cell parameters. Nevertheless, there are still some drawbacks for genetic algorithm, such as low speed.…”