An inverter fault diagnosis method based on ISMO and average log-likelihood ratio

Abstract Conventional methods for diagnosing open-circuit faults in three-phase voltage-source inverters (TP-VSI) suffer from limited diagnostic diversity and slow response times. Addressing this issue, this paper proposes a fault diagnosis method based on an improved sliding mode observer (ISMO). Initially, the construction of the SMO is grounded in the mathematical model of TP-VSI. Additionally, the integration of the power switching function into the SMO aims to mitigate jitter. Subsequently, a novel fault feature is extracted, leveraging the average log-likelihood ratio based on the observer-estimated current and the actual current. Finally, a delayed switching adaptive threshold (DSAT) is devised using logical operations. The DSAT aims to fulfill the requirements for diagnosing two IGBT faults while enhancing adaptability to dynamic operating conditions. Experimental results demonstrate that the proposed fault diagnosis method can effectively identify 21 types of open-circuit faults, achieving a diagnosis time of less than 75% of the current cycle. The method exhibits high accuracy and robustness.