Abstract: In this paper, an intelligent tire algorithm based on support vector machine (SVM) is proposed to predict and classify tire-road friction coefficient. Firstly, the intelligent tire hardware system is developed based on MEMS three-dimensional accelerometer and field test is carried out on three types of road with different adhesion conditions. The radial and lateral acceleration signals are obtained and statistical features in time and frequency domain are extracted. Then, the feature dimension is reduced by principal component analysis (PCA). Based on the feature parameters after dimension reduction, support vector machine is applied for classification training. Finally, the SVM classifier with optimized parameters is used to identify the peak adhesion coefficient. Vehicle field test results show that the proposed algorithm can realize quick estimation of road state, thus providing the key information of road for vehicle control system. Compared with the traditional identification algorithm of adhesion coefficient, the proposed method in this paper is more direct, stable and reliable which does not need acceleration, braking or steering conditions. The method has strong generalization ability, wide application range which is of great potential engineering value

Key words: intelligent tire, tire-road friction coefficient, road identification, principal component analysis, support vector machine