Fault Classification based on Deep Learning for Photovoltaic Systems

In general, high power electronic converters such as Multilevel Inverters (MLI), that find their use in many applications, create difficulties of augmented switch count, capacitor flow current and reverse current flowing from inductive loads. above problems, in turn, result in faults, without which MLI reliability is unattainable. Therefore, fault classification and tolerance strategies are of paramount contribute to keeping reliability of inverters. Herein we present a deep learning (DL)-based fault classification approach for 15-level CHB Multi-level Inverter. proposed system includes pre-processing based on Adaptive Hilbert-Huang filtering, segmentation by a Partial Differential Equation (PDE), feature extraction utilizing Scale-Invariant Feature Transform (SIFT), and fault signal classification being performed by a Convolutional Neural Network (CNN) trained with a Honey Badger Optimization Algorithm (HBOA). MATLAB was applied to our method of identifying a set of four complex fault types for purpose of measurement. results of HBOA-optimized CNN have demonstrated highest accuracy rate of 99.91%.