Improved Fault Classification and Localization in Power Transmission Networks Using VAE-Generated Synthetic Data and Machine Learning Algorithms

• DOI: 10.3390/machines11100963
• Publication Date: 2023-10-16T12:32:57Z
• Authors: Muhammad Amir Khan; Bilal Asad; Toomas Vaimann; Ants Kallaste; Raimondas Pomarnacki; Van Khang Hyunh
• Keywords: Boosting

The reliable operation of power transmission networks depends on the timely detection and localization faults. Fault classification in electricity can be challenging because complicated dynamic nature system. In recent years, a variety machine learning (ML) deep algorithms (DL) have found applications enhancement fault identification within networks. Yet, efficacy these ML architectures is profoundly dependent upon abundance quality training data. This intellectual explanation introduces an innovative strategy for pinpointing faults achieved through utilization variational autoencoders (VAEs) to generate synthetic data, which turn harnessed conjunction with algorithms. approach encompasses augmentation available dataset by infusing it synthetically generated instances, contributing more robust proficient recognition categorization Specifically, we train VAE set real-world data that capture statistical properties To overcome difficulty diagnosis methodology three-phase high voltage networks, categorical boosting (Cat-Boost) algorithm proposed this work. other standard recommended study, including Support Vector Machine (SVM), Decision Trees (DT), Random Forest (RF), K-Nearest Neighbors (KNN), utilizing customized version forward feature selection (FFS), were trained using VAE. results indicate exceptional performance, surpassing current state-of-the-art techniques, tasks localization. Notably, our achieves remarkable 99% accuracy extremely low mean absolute error (MAE) 0.2 These outcomes represent notable advancement compared most effective existing baseline methods.