Improved Fault Prediction using Hybrid Machine Learning Techniques

Due to increasing demand for consumption of electrical power along with the difficulties in expansion of available networks for transmission. Considerably, transmission line is the most relevant part of the power system. The requirement of power along with its allegiance has observed to be exponentially growing over the advanced technical era and the key objective of a transmission line is to pass the electric power from the source to destination of distribution network. The term fault analysis is very challenging in power system engineering to deduct the fault in short time from transmission line as well as re-establish the power system as earlier as possible on very less interruption. The main aim for this study is that fault detection and diagnostics for preventing the loss of electricity is still a key issue of research, and the problem has yet to be solved. Thus, utilizing an intelligent control switch such as the IEC-61850 (International Electro Technical Commission) based on the GOOSE (Generic Object Oriented Substation Event) protocol, a real-time modelling and testing of transmission line error protection and communication is designed. Because transmission line error cannot be avoided in an electrical power system, we employ the GOOSE protocol for communication to convey the detected fault in the transmission line via the remote protection relay. The simulation result is performed by using SVM to train the system and ANN is utilized to classify the occurrence of faults in different types in order to get the satisfactory outcome.