基于电池外特征的粒子群神经网络电池健康状态预测  

Prediction of State of Health Based on Particle Swarm Neural Network with Battery External Characteristics

作　　者：郑永飞 文怀兴[1] 韩昉[1] 杨鑫 ZHENG Yong-fei;WEN Huai-xing;HAN Fang;YANG Xin(College of mechanical and electrical engineering,Shaanxi University of Science and Technology,Xi'an 710021,China;Xi'an GuanTong Digital Source Electronics co.LTD,Xi'an 710065,China)

机构地区：[1]陕西科技大学机电工程学院,西安710021 [2]西安冠通数源电子有限公司,西安710065

出　　处：《科学技术与工程》2019年第36期184-189,共6页Science Technology and Engineering

基　　金：咸阳市二〇一八年科学技术研究计划(2018K02-16)资助

摘　　要：为了降低电池特征参数获取难度,提高电池健康状态(state of health,SOH)预测精度,保障电动汽车安全行驶,针对电池使用过程中内部参数变化复杂难以测量及BP神经网络容易陷入局部最小值等问题,提出了一种基于电池外特征的粒子群神经网络SOH预测方法。将电池的外特征参数电压与温度作为输入,在BP网络的架构中引入粒子群算法对网络的权值与阈值进行优化,从而增强网络的全局寻优能力。在MATLAB 2018上进行仿真验证,实验结果表明,本方法比传统的BP网络适用性更好,精度更高,绝对误差在1.6%以内,相对误差在2.4%以内,具有更广的应用前景。In order to reduce the difficulty of obtaining battery characteristic parameters,improve the accuracy of SOH prediction,and ensure the safe driving of electric vehicles,a particle swarm optimization neural network prediction method of SOH based on external characteristics of batteries is proposed to solve the problems such as the complexity and difficulty of measuring the internal parameters during the use of batteries and the BP neural network falling into the local minimum value.By taking voltage and temperature as input,particle swarm optimization was introduced into the architecture of BP network to optimize the weights and thresholds of the network,so as to enhance the global optimization ability of the network.MATLAB 2018 was used for simulation verification.The experimental results show that this method has better applicability and higher accuracy than the traditional BP network,and the absolute error is within 1.6%and the relative error is within 2.4%,which has a broader application prospect.

关 键 词：粒子群算法 BP神经网络 动力电池 健康状态 

分 类 号：TM912.9[电气工程—电力电子与电力传动] U469.72[机械工程—车辆工程]