为了更加高效地评估储能电池组的健康状态（SOH），提出一种基于电压极差特征的早期健康状态检测方法。首先基于大容量磷酸铁锂储能电池组开展循环老化试验，测量每次循环的电压极差信号，并从中提取关键时间点的电压特征；其次，基于皮尔逊（Pearson）相关系数及灰色关联度分析法（GRA）筛选与电池组老化高度相关的健康因子。最后，通过麻雀搜索算法（SSA）优化双向长短时记忆网络（BiLSTM）的超参数，搭建SSA-BiLSTM健康状态估计模型，实现储能电池组SOH评估；并结合常规机器学习算法验证了健康因子的有效性和估计模型的优越性。结果表明，所提取充放电静置30 min的电压极差特征能够有效反映电池组容量衰退趋势，多种模型验证下SOH估计误差均低于±0.8%。其中，本文所提出的SSA-BiLSTM模型均方根误差（RMSE）低至0.07%。因此该方法能够有效地对大容量储能电池组的SOH实现在线监测。

In order to evaluate the state of health (SOH) of energy storage battery packs more efficiently, an early health status detection method based on voltage range characteristics is proposed. Firstly, the cyclic aging experiment is carried out based on the large-capacity lithium iron phosphate battery pack , and the voltage range signal of each cycle is measured , and the voltage characteristics of key time points are extracted from it. Secondly, health factors highly related to battery aging are screened based on Pearson correlation coefficient and grey correlation degree analysis (GRA) . Finally, the Sparrow search algorithm (SSA) is used to optimize Bi-bidirectional long short-term memory ( BiLSTM ) hyperparameters, and SSA-BiLSTM health state estimation model is built, and realize SOH evaluation of energy storage battery pack. The effectiveness of the health factor and the superiority of the estimation model are verified by the conventional machine learning algorithm. The results show that the extracted voltage range characteristics of charging and discharging for 30 min can effectively reflect the decline trend of battery pack capacity, and the estimated error of SOH is less than士0.8% under various models. The root mean square error (RMSE) of the SSA-BiLSTM model proposed in this paper is as low as 0. 07% . Therefore, this method can effectively monitor the SOH of large-capacity energy storage battery packs online.

国家自然科学基金项目(51577050); 国网江苏省电力有限公司重点科技项目(J2022158)