An iterative model of the generalized Cauchy process for predicting the remaining useful life of lithium-ion batteries. (January 2022)

• Record Type: Journal Article
• Authors: Hong, Guangxu; Song, Wanqing; Gao, Yan; Zio, Enrico; Kudreyko, Aleksey
• Is Part Of: Measurement. Volume 187(2022)
• Publication Date: 2022-01
• Subjects: Lithium-ion batteries -- Remaining useful life -- Long-range dependence -- Generalized Cauchy iterative model -- Predictive maintenance; LRD long-range dependence -- GC generalized Cauchy -- RUL remaining useful life -- ACF autocorrelation function -- FBM fractional Brownian motion -- LSTM long-short-term memory -- FT failure threshold -- MLE maximum likelihood estimation -- PDF probability density function -- RS rescaled range -- FPT first prediction time -- EOL end-of-life -- MAE mean absolute error -- RMSE root mean square error -- HD health degree; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2021.110269
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: The long-range dependent is described by Hurst exponent and fractal dimension. The degradation process of the capacity proved to have long-range dependent. The generalized Cauchy is used as diffusion item to describe long-range dependent. The generalized Cauchy increment is developed by its autocorrelation function. The parameters are obtained by the maximum likelihood estimation method. Abstract: The degradation process of lithium-ion batteries has memory, i.e. it has long-range dependence (LRD). In this paper, an iterative model of the generalized Cauchy (GC) process with LRD characteristics is proposed for the remaining useful life (RUL) prediction of lithium-ion batteries. The GC process uses two independent parameters, fractal dimension and Hurst exponent, to measure the LRD of the degradation process. The diffusion term of the GC iterative model is replaced by the increment of the GC time sequences, constructed via the autocorrelation function (ACF) to describe uncertainty and the LRD characteristics of the lithium-ion batteries capacity degradation. Linear and nonlinear drift terms are used to explain the degradation trend of the lithium-ion batteries capacity. A comparison is made with fractional Brownian motion (FBM) and long-short-term memory (LSTM) network models to show how the GC iterative model has the best performance in RUL prediction of lithium-ion batteries.