Fault Detection in Transmission Lines via Transfer Functions Based on Multi-Agent Reinforcement Learning

In recent years, fault detection addresses high-impedance faults in transmission lines which is crucial for creating reliability and safety in distributed system. The improvements in power system fault detection have employed intelligent techniques like Deep Learning (DL) and Reinforcement Learning (RL) models to improve accuracy and reliability. The existing methods struggled with exact fault detection under varying fault conditions such as changing network topologies, high impedance, and power swings. This research proposed a Multi-Agent Reinforcement Learning (MARL) framework integrated with Transfer Function (TF) analysis to convert the voltage and current signals into the frequency domain, capturing changes caused by faults. The features are extracted through Convolution Neural Network (CNN) to learn automatically and extract complex patterns from TF, and effective detection of faults. MARL employs multiple agents distributed across the power network for detecting and localizing faults through independent learning and decision-making, improving accuracy, adaptability in fault detection tasks. Finally, results of proposed MARL frame work shows that fault detection achieved high performance better outcomes compared to existing Conditional Tabular Generative Adversarial Network (CTGAN) in terms of precision, recall and F1 score by achieving 0.95%, 0.85% and 0.88% respectively.