基于GTO-SVM的轮轨接触预测模型OA

To address the issues of hyperparameter sensitivity and insufficient generaliza-tion ability of the traditional Support Vector Machine(SVM)model in wheel-rail contact state prediction,the Gorilla Troops Optimizer(GTO)algorithm was utilized to globally optimize key hyperparameters of the SVM model,such as the penalty coefficient and ker-nel function,and the GTO-SVM wheel-rail contact state prediction model.was estab-lished.A complete prediction framework encompassing data preprocessing,feature engi-neering,and model validation was constructed,measured multi-source data(encompass-ing wheel-rail contact force,contact position,and wear extent)was acquired from a spe-cific high-speed railway line for the purpose of building an experimental dataset.Simula-tion results show that compared with the standard SVM model,the GTO-SVM model a-chieves a reduced mean absolute error(MAE)in contact force prediction,with the coeffi-cient of determination(R2)improved to 0.996 21.In complex working condition tests,RMSE is lower than that of the traditional model,verifying the strong generalization abili-ty of the GTO-SVM model under complex operating conditions.