Research on Temperature Prediction Model of Thermal Management System Based on BP Neural Network

The thermal hysteresis and inertia slow leads to the slow response speed and high energy consumption of the hybrid vehicle thermal management system, which is a highly nonlinear multi-input and multi-output complex coupling problem. A temperature prediction model for hybrid vehicle thermal management system based on BP neural network prediction is constructed in this paper. The model was trained by changing the learning rate, hidden layers, and hidden layer nodes parameters, and verified and tested the prediction accuracy. The model correlation coefficient R, determination coefficient R2, mean square error (MSE), and mean absolute percentage error (MAPE) were obtained. The comprehensive correlation coefficient R of the predicted model is 0.86. The multi output mean absolute percentage error (MAPEm) of the predicted model is 2.14%, which can be used to provide forward-looking information on the inlet and outlet temperatures of key components of the thermal management system, to lay the foundation for efficient control of the thermal management controller.