基于神经网络的电池SOC估算及优化方法  

Battery SOC Estimation and Optimization Method Based on Neural Network

作　　者：李永颖 张振东[1] 朱顺良 Li Yongying;Zhang Zhendong;Zhu Shunliang(College of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai200093,China;National Motor Vehicle Product Quality Inspection and Supervision Center,New Energy Research Institute,Shanghai201800,China)

机构地区：[1]上海理工大学,机械工程学院,上海200093 [2]国家机动车产品质量检测与监督中心,新能源研究所,上海201800

出　　处：《计算机测量与控制》2020年第5期185-189,194,共6页Computer Measurement &Control

摘　　要：鉴于锂电池高度非线性和时变性使其剩余电量难以精确估算,影响电池的管理和控制;基于BP神经网络模型,在具有随机噪声干扰下,分析和比较不同架构的深度学习模型对电池剩余电量估算的运算时间和泛化性能,并根据粒子群算法(PSO)、基于Nesterov动量的RMSProp变学习率算法优化模型,结合数学规划设计出不同深度的最优构架,并与多种神经网络模型进行比较;根据实验数据和模型估算结果对比表明:此优化算法能有效减少模型的运算时间,在双隐层最优构架下,SOC平均估算误差在0.1左右。In view of the high nonlinearity and time-varying of lithium batteries,it is difficult to accurately estimate the remaining power,which affects the management and control of the battery.Based on the BP neural network model,under the random noise interference,the computational time and generalization performance of the battery's remaining power estimation are analyzed and compared with the deep learning model of different architectures,and based on particle swarm optimization(PSO),Nesterov momentum-based RMSProp,the learning rate algorithm optimization model,combined with mathematical programming,designs the optimal framework at different depths and compared with a variety of neural network models.The comparison between the experimental data and the model estimation results shows that the optimization algorithm can effectively reduce the computation time of the model.Under the optimal framework of the double hidden layer,the SOC average estimation error is around 0.1.

关 键 词：锂离子电池 SOC 神经网络 粒子群算法 RMSProp 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]