AI-Based Predictive Model for Fault Detection in Power Systems Using Radial Basis Function and SVM

Unexpected breakdowns can occur in electrical power systems for a variety of reasons. In order to stop power system problems from spreading to other parts of the protective system, it is necessary to anticipate when they may occur. The protective systems' roles are to identify, categorise, and pinpoint the precise location of the problematic voltage and/or current lines. Finally, the protective relay must communicate with the circuit breaker in order to isolate the problematic line. There are two critical steps in designing the system before training that improve its performance: data preprocessing and feature extraction. The former pertains to the preparation of raw signals for modelling, whereas the latter involves the selection of hyperparameters for the model utilising the SSA. Furthermore, the proposed system is validated utilising two prominent detection models, specifically our four models, which are tested and trained on the publically accessible Kaggle dataset. The experimental findings and discussion reveal that the proposed approach enhances performance in identifying electrical faults with an accuracy of 98.60.