Application of support vector machine in driving ranges prediction of pure electric vehicle with dual – energy storage system based on particle swarm algorithm

Shuang Du^{1, 2}, Chuncheng Zuo¹ 

¹College of Mechanical Science and Engineering, Jilin University, Changchun City, Jilin Province, China

²College of Engineering Technology, Jilin Agricultural University, Changchun City, Jilin Province, China

Driving ranges was a key factor that may affect the popularization and development of pure electric vehicle (PEV) with dual-energy storage system (DESS). It relied on neural network for its prediction. However, the prediction effect was not satisfactory due to local minimization, slow convergence rate, overfitting phenomenon and so on. In order to be more accurate in prediction, this paper introduced the Support Vector Regress (SVR) to the vehicle with parameters optimized by particle swarm optimization (PSO). Compare to BP neural network algorithm, PSO-SVR algorithm is more accurate and practical.