采用集成深度森林模型实现退役电池容量估算

Capacity estimation of retired batteries by fusing ensemble deep forest model

针对退役锂离子电池容量估算面临历史数据缺失,传统机器学习算法存在过拟合和单个模型估算不稳定的问题,提出一种基于集成深度森林的容量估算模型。首先,从退役电池一次满充数据中提取恒流充电时间和充电电流面积特征;然后,利用提取的特征和容量训练多个深度森林建立集成深度森林模型,并设计一种可信状态决策剔除集成模型中波动较大的估算值,取剩余估算值平均值作为最终估算结果。采用自测和公开数据集对所提方法进行验证,结果表明,该方法能实现退役电池剩余容量的准确稳定估算,最大误差仅为0.08 Ah,与传统机器学习算法相比,该方法能获得更高的容量估算精度。

The inability to easily gather historical data,the propensity of traditional machine learning algorithms for overfitting,and the instability of single model estimation make it difficult to estimate retired battery capacity.An ensemble deep forest model was suggested as a solution to these problems.Firstly,the constant current charging time and charging current area features were extracted from the one-time full charging data of retired batteries.Next,several deep forests were trained to create an ensemble deep forest model using the extracted features and capacities.Finally,a plausible state decision was created to exclude the fluctuating estimation values in the ensemble model and took the average value of the remaining estimation values as the financial outcome.The proposed method was validated using self-tests and publicly available datasets.The results show that the method can achieve accurate and stable estimation of the remaining capacity of retired batteries,with a maximum error of only 0.08 Ah,and that it achieves a higher capacity estimation accuracy compared to traditional machine learning algorithms.