Abstract: Taking a three-phase SPWM inverter system as an example, this paper proposes a novel fault diagnosis method for power electronic circuits based on wavelet analysis and neural networks, addressing the limitation of accurate mathematical models for fault diagnosis due to the nonlinear characteristics of power electronic circuits. A nonlinear MATLAB simulation model of the three-phase SPWM inverter power supply is established. The three-phase output fault voltage is used as fault information. Low-frequency energy values ​​are extracted as feature vectors using wavelet analysis. A neural network is used to locate the faulty bridge arm of the inverter. Finally, utilizing the symmetry of the fault voltage of the same bridge arm in the three-phase inverter voltage, a simple judgment logic is used to separate the faulty components. Simulation results show that the method is accurate and reliable. Keywords: Inverter; Wavelet analysis; Neural network; Fault diagnosis In power electronic circuits, most faults manifest as thyristor damage, with open circuits and short circuits being the most common. Unlike analog circuit faults, power electronic circuit faults occur within a very short timeframe, making it difficult for operators to pinpoint the location of the faulty component. Even experienced personnel may be misdiagnosed due to external factors. Therefore, online real-time monitoring and fault diagnosis of power electronic circuits are essential. In recent years, many scholars both domestically and internationally have conducted effective work on this topic. Literature introduces simulation analysis and model-based fault diagnosis methods, but these methods have the drawback of requiring an accurate mathematical model of the system; inaccurate models directly affect the accuracy of the diagnosis. Literature proposes a knowledge-based fault diagnosis method based on current trajectory analysis after inverter component faults. While this method can achieve fault detection and diagnosis, strong external interference can severely affect the current trajectory, hindering fault isolation. For details, please click: A Novel Fault Diagnosis Method for Power Electronic Circuits.