Open AccessOptimization and control of fuel cell thermal management system based on neural network
Author(s) -
Kunhao Tang,
Sanhua Zhang,
Youlong Wu
Publication year - 2021
Publication title -
thermal science/thermal science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.339
H-Index - 43
eISSN - 2334-7163
pISSN - 0354-9836
DOI - 10.2298/tsci2104933t
Subject(s) - direct methanol fuel cell , artificial neural network , computer science , controller (irrigation) , battery (electricity) , voltage , methanol fuel , energy management , thermal , control engineering , automotive engineering , control theory (sociology) , fuel cells , energy (signal processing) , control (management) , artificial intelligence , engineering , power (physics) , chemical engineering , electrical engineering , chemistry , mathematics , biology , anode , quantum mechanics , agronomy , statistics , physics , electrode , meteorology
Aiming at the direct methanol fuel cell system is too complicated, difficult to model, and the thermal management system needs to be optimized. The article attempts to bypass the internal complexity of direct methanol fuel cell, based on experimental data, use neural networks to approximate arbitrarily complex non-linear functions ability to apply neural network identification methods to direct methanol fuel cell, a highly non-linear thermal management system optimization modelling. The paper uses 1000 sets of battery voltage and current density experimental data as training samples and uses an improved back propagation neural network to establish a battery voltage-current density dynamic response model at different temperatures. The simulation results show that this method is feasible, and the established model has high accuracy. It makes it possible to design the real-time controller of the direct methanol fuel cell and optimize the thermal energy manage?ment system?s efficiency.