SVM和Kalman滤波大功率动力电池SOC预估方法的研究

摘  要：新能源汽车大功率动力电池荷电状态（state-of-charge，SOC）的快速精确估计是动力电池能耗管理系统的核心技术，针对大功率动力电池这一非线性、强耦合系统，提出基于支持向量机（support-vector-machine，SVM）静态预测和基于卡尔曼滤波（Kalman）动态预测的动力电池SOC预估方法。仿真实验结果表明，采用基于SVM和Kalman滤波结合的预估方法可以快算完成动力电池SOC的估计，并且动力电池模型参数的变动几乎不影响算法的准确性，表明算法具有一定的鲁棒性。
关键词：荷电状态；卡尔曼滤波；支持向量机；预测模型
中图分类号：U473.4；U461.2；TP391.9；TP301.6       文献标志码：A      文章编号：1674-5124（2013）05-0092-04
Estimation method research of high volume battery based on SVM and Kalman filter
LI Zheng-guo1， MENG Fan-kun2
（1. Shenzhen Polytechnic，Shenzhen 518055，China；
 2. College of Electrical Engineering，South China University，Hengyang 421001，China）
Abstract: Fast and accurate state-of-charge（SOC） estimation about the high volume battery of new energy vehicles is a key technique for energy efficiency management. For the non-linear and inherent dynamic property of a high volume battery， this paper put forward an algorithm for battery SOC static estimation based on support-vector-machine（SVM） and dynamic estimation based on Kalman filter. The simulation result shows that the algorithm can estimate SOC quickly and accurately， and the disturbance of battery model parameters does not influence the accuracy of this method， which shows the robustness of this method.
Key words: state-of-charge； kalman filter； support-vector-machine； predication model