计算机工程 ›› 2012, Vol. 38 ›› Issue (12): 143-145.doi: 10.3969/j.issn.1000-3428.2012.12.042

• 人工智能及识别技术 • 上一篇    下一篇

粒子群优化神经网络在SOC估算中的应用

刘秋丽,马晓军,袁 东,苏建强   

  1. (装甲兵工程学院控制工程系,北京 100072)
  • 收稿日期:2011-08-08 出版日期:2012-06-20 发布日期:2012-06-20
  • 作者简介:刘秋丽(1976-),女,讲师、博士研究生,主研方向:电动汽车技术;马晓军,教授、博士生导师;袁 东、苏建强,博士研究生

Application of Particle Swarm Optimization Neural Network in State of Charge Estimation

LIU Qiu-li, MA Xiao-jun, YUAN Dong, SU Jian-qiang   

  1. (Department of Control Engineering, Academy of Armored Forces Engineering, Beijing 100072, China)
  • Received:2011-08-08 Online:2012-06-20 Published:2012-06-20

摘要: 针对电传动车辆用动力电池组荷电状态(SOC)非线性强、普通神经网络模型预测精度低的问题,提出利用粒子群优化神经网络权值和阈值的预测方法,建立基于该方法的BP神经网络电池SOC训练模型。为克服粒子群算法容易陷入局部最优的缺点,用混沌变量初始化粒子位置,采用可避免粒子高度聚集的算法,提高模型的预测精度。仿真结果表明,使用该方法估算电池的SOC更具快速性、准确性和稳定性。

关键词: 神经网络, 粒子群优化, 荷电状态, 局部最优, 混沌变量, Logistic映射

Abstract: Since the State of Charge(SOC) of electric vehicles battery has the characters of nonlinearity concerned with multifactor, and precision forecasted is difficult to be satisfied by the common artificial neural networks. A novel battery SOC forecast modeling is established by Back Propagation(BP) neural network based on Particle Swarm Optimization(PSO) which is used to optimize the weight and threshold of neural network. For overcoming the shortcoming of PSO algorithm trapped in local optimum, chaotic variables are provided to initialize the position of particles and the algorithm is embedded to avoid massive particles. It is proved through the simulation that the battery SOC is estimated rapidly, precisely and stably by this method.

Key words: neural network, Particle Swarm Optimization(PSO), State of Charge(SOC), local optimum, chaotic variable, Logistic mapping

中图分类号: