Study On The Low Power Proton Exchange Membrane Fuel Cell And Its Control

Nowadays, power crisis and environment pollutions are two critical problems that should be solved urgently.The fuel cells is a kind of high efficiency power supply, which directly turns chemical power into electricity power, without producing any pollution.Among all the different kinds of fuel cells, the Proton Exchange Membrane Fuel Cells(PEMFC) is the most popular one, owing to its advantages of low operating temperature, high power density and fast transient response. But fuel-cell technology is the intersection of kinds of subjects related, which makes it difficult to model the PEMFC. Besides, limited by the materials used, the PEMFC is strict with the operating temperature and humidity.which means that much consideration should be paid to both the design and the control of the PEMFC power supply system. In this paper, research works were conducted in the field of PEMFC test, operation condition optimization, modelling and control strategy, which mainly consist of:(1) Important principles of PEMFC and PEMFC systems were introduced. A preliminary design of the small PEMFC power source was proposed.(2) Experiments were conducted to test the performance of a lab-made PEMFC and optimize its operation conditions. Based on these experiments, optimization was achieved.(3) The modelling of PEMFC is related to kinds of subjects like electrical science, thermodynamics,electrochemistry, et al. Thus, the modelling of PEMFC is difficult and owns inaccuracies. The BP neural network method is a superior alternate to modelling a nonlinear system as complicated as the PEMFC. And the simulation results reveals that the BP neural network model can fitted the experiment data in a good manner, which meets the need of on line control.(4) Under MATLAB/simulink environment, a BP neural network model of PEMFC was established, and the maximum efficiency tracing controllers adopting adaptive sinusoidal perturbation and sliding mode extremum seeking algorithm were designed. The simulation results showed that both the two kinds of extremum seeking algorithm could trace the maximum efficiency point under different work conditions and improve the performance, while sliding mode extremum seeking algorithm controller was superior in robustness and steady-state characteristics.