Enhanced Fault Detection and Localization in Power Transmission Networks Using Deep Learning-Based IoT Technology and Improved Artificial Humming Bird Algorithm (IAHA)

Power transmission networks are very inclined to faults that are difficult to find and fix in real time. This causes more power loss, less reliability, and more safety risks. According to changing operating conditions, current methods often fail to perform well enough to get both high accuracy and fast response at the same time. To tackle these issues, this paper suggests an IoT framework that uses deep learning to quickly detect and localize faults. The suggested system uses a Bipartite Graph Neural Network (BGNN) to find faults accurately and an Improved Artificial Hummingbird Algorithm (IAHA) to tune parameters perfectly. The IoT-based architecture, which includes current-sensing devices, edge computing, and cloud analytics, makes it possible to monitor and make decisions in real time. The proposed BGNN-IAHA model works better than traditional methods like LSTM, MLP, and CNN-LSTM. It has a detection accuracy of 98%, a localization time of 50 ms, and a power consumption of 20 W. The system also supports real-time alerts and GPS-based fault location, which improves safety and response time.