基于GBDT模型预测锂离子电池容量衰减

Predict Lithium-ion Battery Capacity Decay Based on GBDT Model

电动汽车电池包的整包容量决定了汽车的行驶里程,因此需要对电池容量衰减进行分析,评估电动汽车的续航能力。本文基于安时积分法计算每一辆电动汽车每次充电情况下电池包的整包容量,采用线性回归对电池包的整包容量衰减进行分析,以容量衰减率作为不同车辆电池劣化速度的评价指标,并基于梯度下降决策树GBDT模型回归车辆的电池容量衰减率,结果表明基学习器的数量达到200时回归的效果达到最优,预测结果的均方误差占最大衰减率的0.45%。

The battery pack capacity of an EV determines the driving range of the vehicle,so it is necessary to analyze the battery capacity attenuation to evaluate the EV’s range ability. This article is based on an ampere-hour integral method to calculate each electric cars per charge capacity battery pack under the whole package,using linear regression analysis of the whole package of battery pack capacity attenuation,capacity attenuation as evaluation index of different vehicle battery degradation speed,and the model based on decision tree GBDT gradient descent to return to the vehicle’s battery capacity attenuation,the results show that the base of learning number reaches 200 return to achieve the optimal effect,predict the results of the mean square error is about 0.45%of the maximum attenuation.