Remaining useful life prediction of lithium-ion battery based on improved gated recurrent unit-generalized Cauchy process

Lithium-ion batteries have become a core technology in the modern energy field due to their wide application in portable electronic devices, electric vehicles, and energy storage systems, and their health state and remaining life prediction are of great significance. The degradation process of lithium-ion batteries has high complexity due to long correlation and local irregularities. To address this problem, this paper proposes an improved gated recurrent unit-Kolmogorov-Arnold network-generalized Cauchy process (GRU-KAN-GC) method for the remaining life prediction of lithium-ion batteries. The method effectively captures the time series features in the degradation data of lithium-ion batteries by introducing a lightweight gated recurrent unit (GRU) network while optimizing the feature extraction capability by combining with the Kolmogorov-Arnold network (KAN) to improve the model performance. The long correlation in the degradation process is described by the Hurst index, while the local irregularity is quantified by the fractal dimension, and the accurate estimation of model parameters is accomplished using the great likelihood estimation method. In the framework of the first hitting time (FHT), an expression for the approximate probability density function (PDF) of the remaining lifetime based on the weak convergence theorem is derived. Through experimental validation on two public battery datasets, the results show that the proposed method has significant advantages in terms of accuracy and efficiency in health state assessment and remaining life prediction.