Detecting and Classifying Faults in Long Transmission Lines with Wavelet Coefficients and Radial Basis Function Neural Networks

Reliability of the electrical system depends on pre- cise fault classification and detection, especially in long lines of transmission. The accuracy and effectiveness of traditional fault detection systems are frequently compromised, requiring sophisticated fault categorization approaches. In WAVELET transform(WT) based fault detection the threshold values de- termines how faults are categorized. The threshold’s value is not universal, therefore not every power systems can use it. These limitations and the issues related to reliability of the power system can be addressed with intelligent-based solutions. In this paper, a technique for fault identification and classification in trans- mission lines is proposed, utilizing Radial Basis Function Neural Networks (RBFNN) and WT. A case study has been performed in MATLAB software, where WT is utilized to extract the fifth level detail components for various fault conditions and used as dataset for developing the RBFNN model. The suggested method outperforms traditional techniques in accurately classifying faults with an accuracy of 99.995% and mean square error(MSE) close to zero. The efficacy of this method in precisely detecting fault type is demonstrated by comparative analysis using MSE metric.