A novel enhanced deep learning-based fault diagnosis approach for cascaded multilevel inverter

Abstract

With the widespread use of power electronic converters like Multilevel Inverters (MLI) in various applications, several challenges arise due to increased switch count, capacitor surge current, and reverse current from inductive loads. These factors lead to the onset of faults, which negatively affect the reliability of MLI. Hence, an efficient fault detection and fault tolerant technique are necessary to ensure the reliability of the inverter. The deep learning-based fault detection approach is proposed for a 15-level Cascaded H-Bridge (CHB) MLI to address this issue. The proposed technique involves pre-processing with an Adaptive Hilbert-Huang filter, segmentation using Partial Differential Equation (PDE), feature extraction using Scale-Invariant Feature Transform (SIFT), and classification of fault signals with a Crow Swarm Algorithm (CSA) optimized Convolution Neural Network (CNN) classifier. To evaluate the effectiveness of the proposed approach in fault classification, four complex fault types were considered using MATLAB. The results show that CSA-optimized CNN offers a remarkable classification accuracy of 99.84%.