A Prediction Model Based on Artificial Neural Network for the Temperature Performance of a Hydrodynamic Retarder in Constant-Torque Braking Process

Excessively high brake temperature of hydrodynamic retarders may lead to fading and failure, resulting in a decrease effectiveness. However, the performance modeling is challenge because non-linear characteristics system. In this study, model based on an artificial neural network constructed predict retarder constant-torque braking process. The developed from back-propagation trained with Levenberg-Marquardt algorithm. Before application network, computational fluid dynamics used obtain controllable region where experimental tests were performed collect data for training validation. linear regression method adopted check quality training. It shown that within 98% accuracy. Furthermore, retarder, which consists thermal balance models, simulated 1500N·m, 2000N·m, 2500N·m processes. simulation results are agreement 2.87% error. proposed can accurately provide theoretical guidance control strategy management.