摘要
为提高锂离子电池荷电状态的预测精度,将粒子群算法引入到径向基神经网络中,建立锂离子电池荷电状态混合估算算法。采用粒子群算法对径向基神经网络隐层节点中心和宽度及连接权值进行优化,降低径向基神经网络参数取值的繁杂度,提高荷电状态预测精度。利用Arbin BT2000多功能蓄电池测试平台,获取到锂离子电池放电数据,进行模拟训练和预测。实验表明:混合算法相对RBF网络具有更好的预测能力,满足荷电状态估算精度误差小于5%的要求,验证了该模型是有效、可行的。
To improve the prediction accuracy of state of charge（SOC） for the lithium-ion battery, a hybrid algorithm for SOC, which introduced particle swarm optimization（PSO） algorithm into the radial basis function neural network,was proposed. PSO was adopted to optimize the nodes center, the width and the connection weights of the hidden layer, and decreased the complexity of the parameter selection in RBF neural network, and then the accuracy of SOC estimation was improved. The lithium-ion battery discharging data, used to simulate train and predict SOC,came from Arbin BT2000 multi-function battery test platform. The experiment shows that the hybrid algorithm,compared with the RBF network, has the better prediction ability, and satisfies the basic requirement that the error of the accuracy of SOC estimation should be less than 5%, which verifies the model is effective and feasible.
出处
《电源技术》
CAS
CSCD
北大核心
2016年第5期982-985,共4页
Chinese Journal of Power Sources
基金
国家自然科学基金资助项目(51247004)
湖北工业大学2012年博士启动基金项目(BSQD12018)
关键词
锂离子电池
粒子群算法
径向基神经网络
荷电状态
权值优化
lithium-ion battery
particle swarm optimization algorithm
radial basis function neural network
state of charge
connecting weight optimizing
