基于SVM的蓄电池容量智能在线检测技术研究

支持向量机(SVM)是一种基于结构风险最小化原理,具有高泛化性能的学习算法.针对铅酸蓄电池容量测试复杂的非线性过程,提出了基于SVM的多输入信息融合技术.利用光纤铅酸蓄电池容量传感器对铅酸蓄电池容量电池进行了多次不同充放电工况条件下的实验,因输入多电压信号与输出电解液浓度存在着非线性关系,而电解液浓度能很好地反映蓄电池容量,利用SVM拟合该复杂的非线性过程,仿真结果显示拟合效果较好,能将误差控制在-0.012～0.015.试验结果表明,传感器输出的信号与蓄电池容量存在着固定的函数关系,理论预测与实际测量结果基本一致,从而表明基于SVM的光纤铅酸蓄电池容量在线智能检测技术的可行性.

Research on Battery Capacity Fiber On-line Intelligent Testing Technology Based on SVM

The support vector machine(SVM) is a learning arithmetic which based on structural risk minimization theory and has highly generalized performance.The paper aiming at the lead-acid battery capacity testing course,proposes a kind of multi-input information fusion technology based on SVM.A fiber lead-acid battery capacity sensor is used to test the battery capacity under different charge and discharge states.Since the input voltage information is related to the output electrolyte concentration,at the same time the concentration is related to the battery capacity very well,SVM is adopted to fit this complex non-linear course.The simulation result turns out good fitting effect which controls the error between-0.012 to 0.015.The testing result indicates that the theory value is almost identical to the actual value.The lead-acid battery capacity fiber on-line intelligent testing technology based on SVM is feasible.