Research on Fault Identification Algorithm for Photovoltaic Grid-Connected Inverter Based on VMD-Wavelet and SVM

With the large-scale application of solar photovoltaic power generation technology, photovoltaic grid-connected inverters, as the core equipment for energy conversion between photovoltaic systems and public power grids, play a crucial role in ensuring the overall benefits of photovoltaic power generation systems. IGBT power switching devices operate under harsh conditions such as high voltage, high current, and high frequency for long periods, resulting in a high probability of open-circuit faults. Moreover, the circuit can still operate during the initial fault period, and external characteristics are not obvious. Traditional diagnostic methods have limited detection accuracy. This paper proposes a fault identification algorithm for photovoltaic grid-connected inverters based on Variational Mode Decomposition (VMD), Wavelet Packet Transform, and Support Vector Machine (SVM). The algorithm uses three-phase output current from the AC side as the fault observation signal, adaptively decomposes the signal into several band-limited modal components using VMD, and determines the optimal modal number using the minimum sample entropy criterion. Three-layer wavelet packet transform is applied to each modal component, and wavelet packet energy of each sub-band is calculated to form a high-dimensional fault feature vector. After normalization, the feature vector is input into a multi-class SVM classifier to complete accurate fault type identification. A simulation model of three-phase two-level photovoltaic grid-connected inverter is built on MATLAB/Simulink platform. 280 sets of sample data are generated for 7 types of working conditions including 6 IGBT single-device open-circuit faults and normal operation. Experimental results show that the proposed method achieves identification accuracy above 98.5% for all fault types, and maintains good robustness and anti-interference ability under low signal-to-noise ratio conditions.