Internal Short Circuit Fault Diagnosis And Capacity Estimation Of Lithium-ion Batteries

In order to improve the safety and stability of the battery system,the battery management system is used for real-time data collection and condition assessment of the battery,to provide decision-making reference and predictive maintenance information for system management and control,among them,the internal short circuit fault diagnosis and capacity estimation is an important part of the battery management system.In this paper,the internal short circuit fault diagnosis method and the capacity estimation method are presented respectively.For internal short circuit fault diagnosis,combined with multi-resolution wavelet denoising and dynamic time warping algorithm,based on the equivalent resistance simulation of internal short circuit experiment to obtain the charging time-charging voltage data of different degrees of internal short circuit,the internal short circuit fault diagnosis method based on multiresolution wavelet denoising and curve similarity degree was proposed.For the estimation of capacity,the improved multi-scale entropy and numerical transformation method were adopted.Based on GB/T 31484-2015 standard,the capacity test experiment was developed.The data of charging and discharging cycles-capacity retention rate and charging time-charging voltage were obtained.The main content of the paper is as follows:(1)Firstly,based on the development and application of lithium-ion batteries,the research status of internal short circuit fault diagnosis and capacity prediction of lithium-ion batteries at home and abroad were introduced,and the basic characteristics of lithium-ion batteries were presented.(2)Secondly,the equivalent resistance simulation internal short circuit experiment scheme was developed and the experiment was carried out to obtain the charging timecharging voltage data of different degrees of internal short circuit of lithium cobalt oxide battery.Based on GB/T 31484-2015 standard,the capacity estimation experimental scheme was developed and the experiment was carried out to obtain the data of charging and discharging cycles-capacity retention rate and charging time-charging voltage of the 18650 battery.(3)Then,the multi-resolution wavelet denoising method was used to denoise the data of the equivalent resistance simulation internal short circuit experiment.The determination of the denoising parameters was transformed into a mixed integer optimization problem with simple constraints.The objective function of the optimization problem was obtained by cross-validation method,and the optimization problem was solved by genetic algorithm.The energy signal and the dynamic time warping algorithm were used to extract the internal short circuit fault features respectively.The results show that the curve similarity fault features extracted by the dynamic time warping algorithm have better robustness and prediction effect,which verifies the effectiveness of the prediction method based on multi-resolution wavelet denoising and curve similarity fault.(4)Finally,the capacity of the lithium-ion battery is predicted.Numerical transformation is put forward based on improved multi-scale entropy prediction method,using the data from the charging time-charging voltage data,using improved multiscale entropy and energy signal to obtain the actual discharge cycles-capacity retention curve approximation curve,using numerical transformation method makes zoom closer fitting,obtain final forecasting results.Based on the least square error principle,the optimal multi-scale entropy parameters and scaling numerical transformation parameters were determined.The voltage data of the whole region and the voltage data of some intervals were used for prediction,by comparing the actual capacity retention rate with the predicted capacity retention rate,the validity of the capacity prediction method based on the improved multi-scale entropy-numerical transformation was verified.