Commercial vehicle disc brake temperature prediction model construction based on machine learning

Abstract

Abstract The process of conducting thermal-mechanical coupling finite element analysis for commercial vehicle disc brakes is both complex and time-consuming. Moreover, real vehicle testing conditions are diverse, and there are limitations in data collection, impeding a comprehensive understanding of the entire braking process. A pioneering approach is proposed, blending thermal-mechanical coupling finite element simulation with machine learning algorithms, to construct an innovative model for predicting brake temperatures. Employing advanced machine learning algorithms: Random Forest, XGBoost, and CatBoost, the study successfully developed a highly accurate model for predicting the temporal temperature variations at special nodes. The model was evaluated against four quantitative benchmarks, revealing that the Random Forest-based model stands out in terms of both accuracy and stability.