相关向量机在蓄电池剩余容量预测中的应用  

pplication of relevance vector machine in battery's remaining capacity prediction

作　　者：乔波强[1] 侯振义[1] 王佑民[1] 

机构地区：[1]空军工程大学电讯工程学院,陕西西安710077

出　　处：《电源技术》2012年第10期1503-1505,1545,共4页Chinese Journal of Power Sources

摘　　要：剩余容量是蓄电池管理控制的重要参数,为了准确预测阀控式密封铅酸(VRLA)蓄电池的剩余容量,提高预测精度,引入了相关向量机方法对蓄电池剩余容量进行预测,并与最小二乘支持向量机模型、遗传BP神经网络模型的预测效果进行了比较。仿真结果表明,该方法降低了预测模型的复杂度,预测精度高,学习泛化能力强,具有一定的应用价值。Remaining capacity is an important parameter for battery＇s administration and control.In order to predict the remaining capacity of VRLA battery accurately and enhance the prediction precision,relevance vector machine was introduced.Compared with LS-SVM and GA-BPNN model,the simulation results show that the proposed method reduce the prediction model＇s complexity and has higher precision of prediction for practical application.

关 键 词：蓄电池剩余容量 相关向量机 贝叶斯理论 回归预测 

分 类 号：TM912[电气工程—电力电子与电力传动]