基于门控循环神经网络的电池健康状态估算

打开文本图片集

关键词：动力电池；电池健康状态；Adam算法；门控循环神经网络 中图分类号：U469.72 文献标志码：A DOI：10.19822/j.cnki.1671-6329.20250009

【Abstract】A batery Stateof Health （SOH） estimation method based on Gated Recurrent Unit （GRU） neural network isproposedtoaddresstheissueoflowaccuracy inestimatingtheSOHofnewenergyvehicle powerbateries.This method extracts multidimensionalinputfeaturesbasedonbatterychargingdata，performsdata cleaningandnormalizationonthe features，and trainsaGRU network toconstructabatery SOHestimation model.Theresults indicate thattheproposed method can achieve an average absolute error of 0.26% inestimatingbattery SOH，whichis1.O4% lowerthan traditional calculationmethods.This methodcanachieveamoreaccurateestimationof batery SOHandcanbeused forevaluating the aging status of electric vehicles.

Key words：Power battery，State of Health （SOH），，Adam algorithm，Gated Recurrent Unit （GRU）

0引言

随着新能源汽车市场占有量的迅速提升，车辆使用过程中的电池性能及老化程度成为用户关注的焦点。（剩余8029字）