| Abstract: |
In this paper, we present the design and empirical testing of an Intelligent Automated Power Distribution System (IAPDS) that leverages Artificial Intelligence (AI), IoT sensors, and control algorithms for optimal electricity distribution through multi-zone urban grids. Traditional power distribution network faces huge technical losses, long fault restoration time and imbalanced power flow, leading to low reliability of the system and rising operation cost. The proposed IAPDS used a real-time hybrid Long Short-Term Memory (LSTM) and Support vector Machine (SVM) model for fault detection and load forecasting, designed the switching decision making with a Deep Q-Network (DQN) [12] with autonomous switching capability. We verified an empirical study on a six-node, five-zone distribution network over six months that show that the IAPDS achieves 98.2% fault detection accuracy, with a 93.5% average energy efficiency and mean fault response time of 18.4 ms. These results correspond to 23.4%, 14.9%, and 92.5% better performance than SCADA-based systems, respectively. Statistical analysis confirms that it is statistically significant at p < 0.05 in every major performance metric. The system also provides an estimated annual savings of INR 16.2 lakhs/distribution zone three and accounts for about 7.8% reduction in technical losses. The results confirm that AI driven automation in current power distribution systems is indeed a practical proposition providing a framework within which Smart Grid can be implemented in these economies at scale (for eg, India). |
