Intelligent fault diagnosis framework of mechatronics systems on high resolution sensory data

A stable electrical power supply relies on efficient fault diagnosis in power systems. The existing fault detection methods suffer from low accuracy and detection delay, hampering prompt rectification of power system faults. Overcoming these limitations requires precise early fault diagnosis, which is crucial for the evolution of power system fault diagnosis. This paper presents an advanced fault diagnosis method, utilising an enhanced convolutional neural network (CNN) with high-resolution digital fault recorder signals. The method targets subtle fault characteristics of power systems, achieving early fault detection, even before five recording cycles. The performance metrics for the method, including accuracy, precision, recall, and F1-score, are documented as 0.9550, 1.0000, 0.9495, and 0.9735, respectively. Additionally, the outcomes of the proposed work demonstrate enhanced performance compared to existing models. The developed automatic detection system accurately identifies faults, facilitating swift power supply restoration. This approach mitigates potential power system fault consequences, enhancing supply stability and economic benefits while reducing government spending.