Research on neural network application for current sensor fault classification and localization in synchronous motor drive

This research focused on the idea of using an artificial neural network-based fault classification system for current sensor (CS) faults. Analysis includes a wound rotor synchronous motor drive system designed according to the idea of fault-tolerant control. Due to the use of the CS fault compensation strategy, the drive system ensures continued functionality, even after a failure has occurred. In the article, the simulation part of the implementation of a CS fault classifier is presented. The neural structure aims to determine the damage category (gain change, offset, saturation, open circuit) and localisation (phase A or B). The developed classifier uses the principal structure of a multilayer perceptron. The assessment of the type of CS fault is based on single samples of the stator phase current signal. The high precision of the classifier that uses the information from the mathematical model in the practical implementation of diagnostic systems for wound-rotor synchronous machines.