最小二乘支持向量机荷电状态估计方法  

ESTIMATION METHOD FOR STATE OF CHARGE BASED ON LEAST SQUARE SUPPORT VECTOR MACHINE

作　　者：盛瀚民[1] 肖建[1] 贾俊波 邓雪松[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031 [2]淡马锡理工学院工程学院,新加坡529757

出　　处：《太阳能学报》2015年第6期1453-1458,共6页Acta Energiae Solaris Sinica

基　　金：国家自然科学基金(51177137)

摘　　要：为解决一般电池模型对于不同健康状况电池泛化性能较差的问题,利用最小二乘支持向量机(LSSVM)的回归原理,通过提取锂电池运行过程中的外部特性构建LSSVM模型。引入粒子群优化算法(PSO)以提高训练效率与模型精度。通过恒流放电实验比较了几种核函数的估计效果,利用交替充、放电实验验证了PSO-LSSVM方法在复杂运行状况下电池荷电状态(SOC)估计的有效性。并与其他估计、优化方法比较,进一步验证方案的优越性。该方法给微电网储能系统的精确、快速估计提供了新的解决方案。State of charge is an important reference quantity in micro-grid energy management system. To solve poor generalization performance problem of general battery model for batteries in different health conditions, the LSSVM model was proposed based on regress principle of Lease Square Support Vector Machine （LSSVM） and extracting external characteristics of operating lithium battery. Particle Swarm Optimization （PSO） algorithm was introduced to improve training efficiency and accuracy of the model. The estimation effects of several kernels were compared through constant discharge test, the alternating charge and discharge experiment was used to verify the effectiveness of proposed method under complex conditions. Moreover, the comparison was also made using other existing methods to prove the superiority of the proposed method. This new model provides a better solution for micro-grid energy management system.

关 键 词：荷电状态 最小二乘支持向量机 粒子群优化 核函数 BP神经网络 

分 类 号：TM912[电气工程—电力电子与电力传动]