Fault Diagnosis of HCSY-MG Microsource Power Abnormal Fluctuation Based on Multi-Feature Fusion CNN

In a Half-bridge Converter Series Y-connection Microgrid (HCSY-MG), each bridge arm is connected with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$N$</tex> Generation modules (GM) in series. The three bridge arms are connected and combined in a star configuration to form a three phase system. To accurately diagnose the system imbalance problem caused by each micro-source fault in HCSY-MG, this paper adopts a Fast Fourier Transform (FFT). In this method, the in circuit signals such as voltage and current in HCSY-MG are taken as input signals for FFT and input to a one-dimensional CNN, and also as input signals for Wavelet Time Frequency Representation (WTFG) and input to a two-dimensional CNN. Finally, the obtained voltage and current signals are enhanced by the multi-head attention mechanism (MHAM). The network can learn information in different feature spaces in parallel, thus improving the accuracy and comprehensiveness of fault diagnosis.