ADG-GWO: Adaptive Dynamic Guided Grey Wolf Optimizer for Intelligent Scheduling of Power Distribution Networks Based on Load Forecasting

Accurate scheduling optimization in modern power distribution networks requires both reliable load forecasting and efficient global optimization. However, existing approaches often suffer from performance bottlenecks such as poor adaptability to fluctuating load patterns, slow convergence under dynamic conditions, and limited coupling between forecasting accuracy and optimization decisions. To address these challenges, this paper proposes an Adaptive Dynamic Guided Grey Wolf Optimizer (ADG-GWO) integrated with a multi-stage load-aware forecasting (MLF) mechanism. The proposed framework first employs a deep learning-based load prediction model to capture temporal and seasonal variations in electricity demand. The predicted load profiles are then incorporated into an enhanced grey wolf optimization process. Specifically, the ADG mechanism dynamically adjusts the leadership hierarchy and explorationexploitation balance based on the wolves adaptive fitness values, enabling the algorithm to efficiently escape local optima while accelerating convergence. Experimental results on several real-world power distribution datasets demonstrate that the proposed ADG-GWO outperforms traditional GWO and other metaheuristic algorithms in terms of convergence speed, scheduling cost, and load balancing stability. Furthermore, the integration of forecasting and optimization offers a proactive scheduling strategy that enhances both reliability and economic efficiency of smart grid operations.