Intelligent diagnosis of cascaded H‐bridge multilevel inverter combining sparse representation and deep convolutional neural networks

Effective fault diagnosis for cascaded H-bridge multilevel inverter (CHMLI) can reduce failure rate and prevent the unscheduled shutdown. Nevertheless, traditional signal-based feature extraction selection methods show poor distinguishability insufficient features in a one-dimensional space. The shallow learning models are prone to fall into local extremum, slow convergence speed overfitting. To cope with these problems, novel image-oriented strategy based on sparse representation (SR) deep convolutional neural network (DCNN) is proposed CHMLI. Initially, Hilbert–Huang transform (HHT) applied obtain HHT spectral images of original monitoring signals, where comprehensively represent detailed information multiple domains time-frequency plane. Furthermore, an image fusion method SR algorithm employed same category construct fused images, which effectively reflects complicated relationships between measured signals features. Ultimately, DCNN not only mine relationship various categories different but also alleviate problem overfitting that caused by limited availability training samples.