A novel hybrid framework for predicting the remaining useful life of energy storage batteries

Abstract

Accurate prediction of the remaining useful life (RUL) of energy storage batteries plays a significant role in ensuring the safe and reliable operation of battery energy storage systems. This paper proposes an RUL prediction framework for energy storage batteries based on INGO-BiLSTM-TPA. First, the battery’s indirect health index is extracted by combining principal component analysis and the Pearson correlation coefficient in the battery charge/discharge cycle data. Second, for the problem that the Northern Goshawk Optimization (NGO) algorithm is prone to falling into local optimum, the Gaussian variation mechanism and nonlinear hunting radius are introduced to improve the NGO algorithm, and the Improved Northern Goshawk Optimization (INGO) algorithm is proposed. Finally, the temporal pattern attention (TPA) mechanism is introduced in the bi-directional long short-term memory (BiLSTM), which makes the model weighted to focus on the features of important time steps, and the INGO algorithm is applied to it to build the RUL prediction framework. Based on the CALCE battery dataset, the root-mean-square error (RMSE) of RUL prediction based on the proposed framework is controlled within 1.3%, which provides better prediction accuracy and generalization.