基于BP神经网络的电池电解液密度预测  

Forecast of density of battery electrolyte based on BP neural network

作　　者：杜灵根 彭澎 

机构地区：[1]武汉船用电力推进装置研究所,湖北武汉430064

出　　处：《电源技术》2017年第11期1605-1607,共3页Chinese Journal of Power Sources

摘　　要：为了对电池电解液密度进行预测,建立了BP神经网络模型,用电池充放电试验数据对其进行了训练和检验。利用训练后的神经网络模型进行了电池电解液密度的预测,预测值与实测值的最大误差值为0.020 9 g/cm3,均方根误差值为0.004 0 g/cm3左右。结果表明,BP神经网络方法可以满足预测精度要求,从而可用于建立电池剩余电量实时监测系统,降低电池维护工作量并延长电池的使用寿命。A BP neural network was built to forecast the density of battery electrolyte. The neural network was trained and tested by data of discharge test. The trained neural network model was used to forecast the density of battery electrolyte, the maximal error of the value of predictions and measurements is 0.020 9 g/cm3, root mean square error is about 0.004 0 g/cm3, which verify that the BP nural network method can meet the demand of the density of battery electrolyte forecast, can be used to establish the observation system of the density of battery electrolyte, and to decrease the workload of battery maintenance and extend the useful life of battery.

关 键 词：电池 BP神经网络 电解液密度 SOC 

分 类 号：TM912[电气工程—电力电子与电力传动]