基于人工神经网络的最大充放电功率预测方法研究  

Prediction Method of the Maximum Charge and Discharge Power Based on Artificial Neural Network

作　　者：李智勇[1] 

机构地区：[1]重庆邮电大学自动化学院模式识别及应用研究所,重庆400065

出　　处：《汽车零部件》2012年第3期50-54,共5页Automobile Parts

摘　　要：针对混合动力汽车中镍氢电池组模块,通过人工神经网络算法预测出其在下一时刻的最大充放电功率值。首先,通过查匹配表获得目标值与SOC(剩余电荷量)、电压、温度的对应关系,再选取与目标值相关联的SOC、电压、电流、温度,加上上一时刻预测出的最大充放电功率值作为该人工神经网络的输入变量,通过实验对人工神经网络结构算法进行设计优化。最终得到最大充放电功率预测用人工神经网络模型,经实际数据测试分析,其误差小于8%。此预测对于混合动力汽车在启动和爬坡控制策略有重要的实用意义。In view of the hybrid electric vehicle＇ s Ni-MH battery module, the next moment of maximum charge and discharge power value was predicted through the algorithm of artificial neural network. Starting from the view of engineering application, checked the matching table to get the target value＇ s relation with the SOC （the rest charge value）, voltage, temperature value, and then selected some of the SOC, voltage, current, temperature which were associated to the target value, plus the last moment of the predicted maximum charging and discharging power value, to be the input variables of the network, and then through the tests, design and optimization the artificial neural network model structure, finally get the artificial neural network model which can successfully predict the target value, and its error within 8% to meet the expected requirements. This prediction for hybrid vehicles on the start-up and climbin~ has the important practical significance.

关 键 词：混合动力汽车 镍氢电池 最大充放电功率 人工神经网络 

分 类 号：U469.72[机械工程—车辆工程]