Modeling method of whole vehicle controllability based on support vector machine and its simulation application

This paper proposes a controllability model based on the support vector machine (SVM) method by analyzing driver controllability evaluation data under abnormal vehicle motor conditions, such as unexpected torque and loss of regenerative torque. First, through controllability tests, vehicle attitude data and full vehicle controllability ratings were extracted to establish a real vehicle database. Then, the Spearman correlation coefficient method and the random forest-recursive elimination method were used to filter the data features. The support vector machine method was applied to explore the mathematical relationship between vehicle attitude data and controllability ratings, providing a controllability model applicable to scenarios such as unexpected acceleration, unexpected deceleration of the subject vehicle, and unexpected deceleration of the lead vehicle. Finally, the controllability model was integrated into a simulation environment, identifying the controllable safety boundaries of the motor at different vehicle speeds, guiding the functional safety development of the entire vehicle.