Research on Neural Networks Identification of a Nonlinear Modeling for PEMFC

The proton exchange membrane fuel cell (PEMFC) as the most promising “green” power technology has the characteristics of the high energy conversion, no electrolyte leakage, low-temperature start. However the operating principles of the PEMFC system comprise a complex nonlinear system. So it is difficult to establish a mathematical model. A method using neural network identification based on Matlab is put forward to research a nonlinear modeling for PEMFC system in this paper. This paper firstly analyzes the necessity of the PEMFC generation technology simply, then introduces the generating principle and utilizes a approach of self-study ability of artificial neural network to build the model of nonlinear system, and adapts the Levenberg-Marquardt BP(LMBP) to build the electric characteristic model of PEMFC. The model and simulation of the system are established in Matlab. By setting the simulation parameter and changing the different cell temperatures, simulation of the different status is done. According to the voltage -current output curves of model, it has some guidance effect for improving the cells performance, and provides basic testimony for optimizing cells performance. It possesses practical significance.