Fault diagnosis method of H-bridge cascade inverter based on data driving

The invention discloses a fault diagnosis method for an H-bridge cascaded inverter based on data driving, and aims to solve the problems that after different power tubes of the cascaded H-bridge multilevel inverter break down, the output voltage/current waveform similarity degree is high, fault diagnosis is difficult, and the actual operation process is affected by direct current side voltage fluctuation, alternating current side load change and circuit noise disturbance. The invention provides a cascade H-bridge inverter fault diagnosis method based on EEMD-MPE (Ensemble Empirical Mode Decomposition-Maximum Power Exchange). The method comprises the following steps of: firstly, carrying out ensemble empirical mode decomposition (EEMD) on voltage of each IGBT transistor after a fault, selecting an optimal intrinsic mode component (IMF) according to the correlation between each component and an original signal, and then calculating a multi-scale permutation entropy (MPE) by using the obtained optimal intrinsic