Incipient Fault Diagnosis of a Grid-Connected T-Type Multilevel Inverter Using Multilayer Perceptron and Walsh Transform

This article deals with fault detection and the classification of incipient intermittent open-transistor faults in grid-connected three-level T-type inverters. Normally, algorithms are developed for permanent faults. Nevertheless, difficulty to detect is much greater, appropriate methods required. requirement due fact that over time, its repetition may lead failures irreversible degradation. Therefore, early these very important ensure reliability system avoid unscheduled stops. For diagnosing faults, a novel method based on Walsh transform combined multilayer perceptron (MLP)-based classifier proposed this paper. non-classical approach using not only allows accurate detections but also fast. last characteristic applications their practical implementation. The includes two main steps. First, acquired AC currents used by control processed transform. results detailed information potentially identify Then, such MLP finally determine whether present or not. Several experiments conducted different types transistor create relevant dataset.