Deep Learning for Real-Time Fault Detection in Power Systems

It is still crucial to the dependability and stability of energy distribution networks to detect the fault in real time. Later on, it will be explained that these more traditional testing methods of defect detection may be less effective, as these methods require predefined thresholds or criteria for comparison, while the system is behavior in complex and dynamic ways. In this research, an approach to real time defects identification of power systems based on deep learning technique is presented. Diagnosing defects is done through using CNN's together with a network that attempts to draw an area of a system that may be problematic, and increasing the accuracy of the identification of defects. However, the previous two methods are significantly outperformed by the proposed CNN method regarding accuracy on the detection of malfunction of the gearbox and the distribution system. This technology is used for the real time monitoring systems, and when the challenges are addressed, the power operators are able to prevent cascade failure or outage in the power chain. Second, this deep learning model is studied with cloud applications and IoT sensor for embedding smart utilities at a larger scale.