Abstract: We constructed underwater equipment applied SOFC system heat management model based on BP network. We then performed linear prediction for the model. We also repeatedly trained the models of the combustor, heat exchanger and pile, selected proper parameters and tested the trained models. The error between tested data and practical data satisfied the requirement. We further applied the same data to the training and test of BP network based SOFC system model. Results show that the error between tested data and practical data is about ±1℃.The model can therefore provide important theory and methodology support for the development of the prototype of SOFC system.

Key words: solid oxide fuel cell, temperature, heat management, BP network model