Centralized Fault Detection and Classification for Motor Power Distribution Centers Utilizing MLP-NN and Stockwell Transform

Industrial processes require continuous operation of low voltage Motor Power Distribution Centers (MPDC). Hence, lowering the interruption rates via pioneering a fault analysis scheme is necessary for the optimum reliability of such systems. This article presents an advanced approach using Artificial Neural Networks (ANN) and Stockwell Transform (ST) to detect and classify Single Line to Ground (SLG) faults in a simulated MPDC. The virtual faulty and healthy three-phase current signals were measured from a centralized position, that represents the accumulation of all load current waveforms, and processed through ST to obtain behavioral characterizing features. Whereas the Multilayer Perceptron Artificial Neural Network (MLP-NN) utilized the statistical features to diagnose faults. The presented results confirm the effectiveness of the proposed fault diagnosis scheme.