Open Circuit and Short Circuit Fault Diagnosis Method for 5-Level Hybrid T-Type ANPC-FC Inverter Using Artificial Neural Networks

Multilevel inverters have been widely used in industrial motor drives and propulsion systems due to their output performance. However, their high number of semiconductor devices increases the risk of failure, potentially causing severe system damage. Being able to localize a fault is the first step in ensuring fault-tolerant operation and such diagnosis method should be quick, simple, reliable and of low implementation cost. This work presents an artificial neural network (ANN) method to diagnose open-circuit and shortcircuit faults in a hybrid five-level active neutral-point clamped (5L-ANPC) inverter.