Experimental Research On Prediction Method For Automotive Suspension Distance And Additional Steer Angle Based On Support Vector Machine

Automotive suspension distance and Steering wheel additional steer angle are the main parameters of vehicle electronic control system, and their observation accuracy influence vehicle dynamics control precision. Based on real car tests data, the paper researched on real time observation method for automotive suspension distance and steering wheel additional steer angle, the main research contents were as follows:(1) Based on the ADAMS/Car software, the paper analyzed the main influencing factors of additional steer angle, they were velocity, pavement amplitude, step steer angle, longitudinal acceleration and deceleration. The results show that lateral acceleration was the main factor that influences the additional steer angle, followed by the longitudinal acceleration, deceleration and pavement amplitude, longitudinal velocity had little effect on additional steer angle.(2) Based on support vector machine (SVM), the paper established suspension distance SVM model. While establishing suspension distance SVM model, taking road condition into consideration. As a result the road condition was employed as input vector of the SVM model.7 kinds of road condition tests and data analyses shown that the controller consumes less than 1 ms when completing wavelet filtering and predicting once in which ε equals 0.01, the squared relationship coefficient of the suspension displacement curves of measured in road way tests and predicted by SVM model were more than 90%. That is to say, the system could meet the requirements of accuracy and real-time control.(3) Based on support vector machine (SVM), the paper established additional steer angle SVM model. The input vectors of additional steer angle SVM model were mainly configuration information of ESP sensors. The model was trained and generalized using the data of typical serpentine tests and FMVSS126 tests. The results indicated that prediction accuracy and real-time performance of the model could meet the requirements of real-time control of vehicle electronic control system.