Optimizing smart nano grid control strategies through virtual environment and hybrid deep learning approaches

Abstract

Smart nano grids face enormous challenges stemming from variations in time demand and risks from cyber threats, which lead to their inefficiency and unstable operation. To overcome these difficulties, this study presents a novel “Federated Reinforced LSTM-Crayfish Whale Optimization Detection (FRLC-WOD)” procedure. The system proposed here integrates Reinforcement-LSTM-Crayfish Optimization Technique (RL-LSTM-CAO) and Federated Graph Whale Optimization Intrusion Detection (FG-WOA-ID) to improve adaptability, efficiency, and security. The RL-LSTM-CAO methodology employs Bi-directional Long Short-Term Memory (Bi-LSTM) for accurate forecasting, Reinforcement Learning-based Power Distribution (RL-PD) for real-time adaptability, and Crayfish Optimization Algorithm (CAO) for optimal energy management. On the other hand, FG-WOA-ID employs Federated Learning for decentralized anomaly detection, Graph Neural Networks for intrusion detection, and Whale Optimization Algorithm for cybersecurity measures adaptation. The results of the experiments achieved a grid stability improvement of 95 %, an energy efficiency improvement of 92 %, a response time of 1.5 s, and a 95 % improved cyber threat resistance, outperforming existing standard methodologies such as EMS GWO-OSA, RNN, and MPPT. This will show how the proposed method significantly upgrades the delivery of reliable and optimized operations for smart nano grids.