Impedance prediction model based on convolutional neural networks methodology for proton exchange membrane fuel cell. (20th May 2021)

• Record Type: Journal Article
• Authors: Ma, Tiancai; Zhang, Zhaoli; Lin, Weikang; Cong, Ming; Yang, Yanbo
• Is Part Of: International journal of hydrogen energy. Volume 46:Number 35(2021)
• Page Start: 18534
• Page End: 18545
• Publication Date: 2021-05-20
• Subjects: Proton exchange membrane fuel cell -- Electrochemical impedance spectroscopy -- Randles equivalent circuit -- Convolutional neural network -- Parameter prediction; Hydrogen as fuel -- Periodicals; Hydrogène (Combustible) -- Périodiques; Hydrogen as fuel; Periodicals; 665.81
• DOI: 10.1016/j.ijhydene.2021.02.204
• Languages: English
• ISSNs: 0360-3199

Abstract

Abstract: Electrochemical impedance spectra is a useful tool for fuel cell water content analysis. However, the data processing efficiency is greatly reduced with the increment in the number and complexity of the experimental parameters. Thus, in this work, an impedance prediction model is developed to assess the impacts of operating conditions and cell position on fuel cell impedance. The model is based on the Randles equivalent circuit and convolutional neural networks. And after verification, this model shows satisfactory performance in impedance prediction. Based on the model, impedance parameters and impedance spectra can be obtained rapidly for any operating condition and cell position. By directly predicting the impedance parameters, a more detailed description about the impact of different operating conditions on internal state of fuel cell can be given. The model can be integrated into the fuel cell control system and determine the optimal operating parameters. Highlights: A CNNs model is used to predict the impedance parameters of the PEMFC. The model can obtain impedance parameters rapidly under any operating condition. The internal state of fuel cells can be analyzed based on the impedance parameters. The optimal operating parameters can be determined more easily.