<p>Faults in high-voltage transmission lines critically impact the stability and reliability of electricity supply, with severe consequences for economic activities and national security. Existing methods face several key issues, such as Limited detection capability, inaccurate classification due to inability to handle complex decision boundaries, and Poor localization accuracy. To resolve these issues, this work develops a novel Deep Learning-based method that effectively detects and classifies faults in high-voltage transmission lines. The proposed approach uniquely combines three advanced techniques, such as Maximal Overlap Discrete Wavelet Transform-based Multi-Resolution Analysis, Bagging Ensemble Sparse Oblique Decision Tree, and Random Effect Bayesian Neural Network. Maximal Overlap Discrete Wavelet Transform-based Multi-Resolution Analysis is employed for accurate fault detection, enhancing the model's sensitivity by extracting significant features from transient signal data. Bagging Ensemble Sparse Oblique Decision Tree is deployed for precise fault classification. It effectively handles complex decision boundaries using oblique splits and ensemble learning. Random Effect Bayesian Neural Network is applied to the fault localization phase, improving localization accuracy by modeling uncertainties in fault data through random effects. The experimental outcomes demonstrated that the proposed integrated methods efficiently predicted, classified, and localized the faults in the transmission line when compared to existing methodologies, underscoring the proposed model’s potential to strengthen the stability and reliability of the power system in real-world applications.</p>