Research on interactive multi‐model fault diagnosis method of Li‐ion battery based on noise suppression

Abstract

The multi-mode-adaptive estimation (MMAE) algorithm lacks the noise suppression capability as impacted by the isolation of each model, which reduces the accuracy of fault diagnosis and causes false alarms. Accordingly, an interactive multiple model (IMM) algorithm was proposed in the present study combined with unscented Kalman filter (UKF) to diagnose multiple faults of lithium-ion batteries. The IMM algorithm consisted of the Markov transition probability matrix (TPM), which enabled the real-time interaction of information at the input of each model. Moreover, the output end fed the updated probability information of the respective model to the input end of the filter by complying with the TPM, which reduced the effect of noise on the algorithm. As demonstrated by the experimental research, the IMM-UKF algorithm is capable of reducing the error attributed to noise, increasing the accuracy of lithium-ion battery fault diagnosis, lowering the fault false alarm rate, and accurately detecting battery fault information.