Modeling of Proton Exchange Membrane Fuel Cell Based on LSTM Neural Network

作者：Zijun Ren, Yigeng Huangfu, Renyou Xie, Rui Ma · 年份：2020 · DOI：10.1109/cac51589.2020.9326514 · 被引用次数：5 · 研究领域：Fuel Cells and Related Materials、Electrocatalysts for Energy Conversion、Advancements in Solid Oxide Fuel Cells

The proton exchange membrane fuel cell(PEMFC) is a complex nonlinear dynamic system coupled with multiple physical fields. To analyze the external characteristics of the fuel cell system, it is regarded as a black box and the neural network structure is used to identify the fuel cell model. This paper uses long and short-term memory(LSTM) neural network to model the PEMFC system, the model's input is the cell current, and the model's output is the cell voltage. First, fit the PEMFC voltage and current characteristic curve under certain operation conditions according to the empirical formula, and train the LSTM neural network through the data obtained above, then continuously optimize the relevant parameters of the LSTM neural network to make the LSTM neural network model output and experience after the training is completed the formula fitting output is the same, and the result shows that the LSTM neural network model can effectively reflect the output characteristics of PEMFC.