摘要
为高效、准确的预测动力电池剩余容量,文中提出了一种基于长短时记忆神经网络的动力电池剩余容量预测方法。首先基于理论推导,明确输出电压、输出电流与剩余容量间存在隐函数关系;然后基于磷酸铁锂电池放电实验的历史数据,通过长短时记忆神经网络搭建预测模型对电池剩余容量进行估计;最后,与常见预测方法进行对比分析。结果表明:文中方法预测误差差值范围在[-0.03,0.06]间分布,平均预测误差为0.016Ah,具有较高的预测精度。同时与其他常用预测方法相比,文中方法精度提高了50%左右,在算法稳定性和识别效率上也有了很大提升。
In order to predict the residual capacity of power battery efficiently and accurately,a prediction method of residual capacity of power battery based on long short term network is proposed.Firstly,based on the theoretical derivation,it is clear that there is an implicit function relationship between the output voltage,output current and residual capacity.Secondly,based on the historical data of lithium iron phosphate battery discharge experiment,a prediction model is built by long short term network to estimate the residual capacity of the battery.Finally,it is compared with the common prediction methods.The results show that:the prediction error difference range of this method is between[-0.03,0.06],and the average prediction error is 0.016 Ah,which has high prediction accuracy.At the same time,compared with other common prediction methods,the accuracy of the proposed method is improved by about 50%,and the algorithm stability and recognition efficiency are also greatly improved.
作者
刘超
李文辉
王克敏
王瑜曈
贠亚玲
LIU Chao;LI Wenhui;WANG Kemin;WANG Yutong;YUN Yaling(State Grid Gansu Electric Power Company Information and Communication Company,Lanzhou 730050,China)
出处
《大电机技术》
2022年第4期70-73,88,共5页
Large Electric Machine and Hydraulic Turbine
基金
国家电网公司科技项目“基于远程监控及智能养护的多站融合储能系统技术研究及应用”(522723190001)。
