A Data-driven Indirect Prediction Of The SOH And RUL Of Lithium-ion Batteries

Lithium-ion battery is a key part of energy storage system,during the continuous charging and discharging process,the performance of lithium-ion batteries deteriorates with decreasing capacity and increasing impedance,which will lead to devices and system failure or even catastrophic loss.Therefore,studying the degradation state of lithium-ion batteries,and then predicting its state of health（SOH）and remaining useful life（RUL）is of great significance to the reliability of the system.However,it is difficult to directly monitor the capacity that characterizes the degradation state of lithium-ion batteries in practical applications.Therefore,this paper analyzes the degradation state of lithium-ion batteries and extracts a health indicator（HI）that can reflect the degradation of capacity to obtain online prediction of SOH and RUL of lithium-ion batteries.In view of the problem that the capacity of lithium-ion batteries is not easy to measure online,a method based on indirect HI extraction is proposed in this paper,which the voltage and temperature information during the charging and discharging process of lithium-ion batteries is analyzed,and extracts and Pearson and Spearman correlation analysis is used to evaluate the relation between HI and capacity;Aiming at the phenomenon of capacity recovery due to the capacity degradation of lithium-ion batteries and the difficulty in measuring the capacity,a new method combining complete ensemble empirical mode decomposition with adaptive noise（CEEMDAN）and support vector regression（SVR）is proposed to predict the RUL of lithium-ion batteries.First,a measurable HI is extracted during the discharge process,and the correlation between HI and capacity is analyzed using the Pearson and Spearman method,then the HI was decomposed by CEEMDAN to obtain a series of relatively stationary components,and finally components are used as the inputs of the SVR model,and the capacity is used as the output to predict the SOH prediction of lithium-ion batteries,which is verified based on the lithium-ion battery degradation data set provided by NASA PCo E,and the experimental results are analyzed,and the RMSE is at least 0.0022.In order to enhance the interpretability and robustness of the experimental results,a multikernel relevance vector machine（MKRVM）model based on probability theory is used to predict the SOH and RUL of lithium-ion batteries in this paper.Comprehensively consider the global and locality of the kernel function to correspondingly solve the trend and volatility in the process of capacity degradation,and improve the prediction ability of MKRVM.In addition,the whale optimization algorithm（WOA）is used to optimize the parameters of the MKRVM model,and finally the model is verified based on experimental data,The experimental results show that the model proposed in this paper has high accuracy.