基于深度学习的储能锂离子电池健康状态估计

Real-time state of health estimation of lithium ion battery in energy storage power station based on deep learning

锂电池储能系统在促进大规模清洁能源并网和保证电网稳定运行等方面发挥着重要作用。针对储能锂离子电池运行过程中的健康管理问题,提出了一种基于深度学习的储能锂离子电池实时健康状态估计方法,利用注意力机制的长短期神经网络,实时评估站内电池单体及电池簇的健康状态。通过对输入特征进行注意力加权,强化参数敏感性较高的特征在模型训练时的作用,以获得更高的估计精度。为验证该模型的有效性,利用公开数据集及实际储能锂离子电池运行数据,分别对储能电池单体及电池簇进行健康状态估计,实现了比传统神经网络方法更高的估计精度。

Lithium ion battery energy storage system plays an important role in stablizing the operation of power grid and promoting the large-scale development of renewable energy.Aiming at addressing the problem of health management during the operation of energy storage power stations,a real-time state of health estimation method for lithium ion bat-teries in energy storage power stations based on deep learning is proposed.A long-term and short-term neural network combined with attention mechanism is used to estimate the real-time state of health of the battery cells or battery clus-ters in power station.This method strengthens the features with higher parameter sensitivity during model training through attention weighting on the input features.Compared with traditional neural network methods,it has higher pre-diction accuracy.In order to verify the effectiveness of the model,a public data set and an actual energy storage power station operation data were used to estimate the state of health of energy storage battery cells and battery clusters,and achieves the forecast effect that the average absolute value error of battery cells and clusters is less than 0.314%and 1.71%,respectively.