Research On Fault Diagnosis Method Of Short Circuit In Lithium Ion Battery Based On VMD And SVM

Lithium ion batteries are widely used in energy storage power stations and electric vehicles due to their high energy density and long service life.However,in recent years,lithium ion batteries have frequently experienced spontaneous combustion and even explosions,resulting in casualties,indicating that the status monitoring and control of lithium ion batteries are facing serious challenges.This paper proposes a method for diagnosing internal short circuit faults in lithium ion batteries based on the combination of variational mode decomposition(VMD)sample entropy and support vector machines(SVM),aiming at the phenomenon that internal short circuit faults in lithium ion batteries cause thermal runaway and lead to spontaneous combustion or even explosion during operation.Firstly,aiming at the problem of non-stationary and nonlinear voltage signals from lithium ion batteries,a signal processing method using VMD to decompose the voltage signals from lithium ion batteries is adopted.Decomposition level K and α penalty factor for two important parameters in VMD,the whale optimization algorithm(WOA)is used to select the optimal K and α.To prove that the optimization algorithm is optimal,the voltage signal of lithium ion batteries is decomposed by VMD in MATLAB.The results show that the optimal parameter combination selected by WOA has a good decomposition effect on the voltage signal of lithium ion batteries.In order to prove that VMD can significantly improve the modal aliasing phenomenon in the signal decomposition process,the empirical mode decomposition(EMD)and ensemble empirical mode decomposition(EEMD)algorithms are compared and analyzed using simulated signals.Secondly,aiming at the difficulty of extracting fault features from lithium ion batteries and the low recognition rate of SVM classification,a fault diagnosis and classification method for lithium ion batteries using sample entropy for fault feature extraction and particle swarm optimization(PSO)algorithm to optimize SVM is proposed.First,the original signal is decomposed into several modal components through VMD,and then the sample entropy of each modal component is calculated to form a feature vector for SVM training and testing.The final output lithium ion battery fault diagnosis rate is 66.7%.After feature selection using PSO for SVM,the fault diagnosis rate is 96.7%,which verifies the superiority of the VMD-PSO-SVM model.Finally,the WOA optimized VMD combined with PSO optimized SVM algorithm model was migrated to the JESTON TX2 platform,and the final diagnostic rate was 96.7%.This verified the feasibility of the diagnostic model and provided a new idea for the internal short circuit fault diagnosis of lithium ion batteries.