Study Of Performance Of Proton Exchange Membrane Fuel Cell Based On Artificial Neural Network

About study of performance of proton exchange membrane fuel cell (PEMFC), theoretical model can reflect the PEMFC output characteristics, and often be used to design battery and analyze operating performance. But the theoretical model was usually built to describe its mechanism by assuming ideal conditions, with a certain degree of regularity, but limited use. Experimental study can directly reflect the characteristics of the battery, but can not reflect the complex phenomena taking place within the battery, and it consumes time and energy. In this paper a new model based on artificial neural network (ANN) was developed combined with experimental data and improved theoretical models, this will not only compensate for the lack of the number of experimental date, but also reduce the errors exist in model only depend on theory by using experimental data.For single-phase model, first of all, build theoretical model according to the PEMFC principle and priori knowledge, optimize parameters in the model by genetic algorithm based on experimental data. Then calculate current density and the corresponding output voltage by the theoretical model, take calculate result and the experimental data as training samples for ANN. This paper gives a three-layer (input layer, hidden layer, output layer) artificial neural network model, with current density, temperature and pressure as model input variables, and output voltage as output variable. There are 142 groups data for training, of which 30 groups are experimental data and 112 groups are calculate value by the theoretical model. Here take 3/4 sample data as neural network training data, then use the other 1/4 data for examination. Trained ANN model can be used to predict the performance of the PEMFC. At high current density, due to the presence of liquid water, the model prediction error increases. When the temperature rises, the performance of the PEMFC increase and then decrease because of the same reason- the presence of liquid water.In this paper, determined the critical current density according to the balance of water at various temperatures, from which can get average liquid saturation, then build a theoretical model when liquid water exist in the PEMFC. Study practical significance using of single-phase and two-phase model at different current density and temperature, the results can describe the state conversion occurred in the PEMFC, and it can make the ANN model prediction closed to the experimental result.The result in this study also indicates that the dynamic behavior occurs in the PEMFC directly affect the performance of the PEMFC, preparing for further study.