A Fault-Tolerant Distributed Agent-Based Framework for Active/Reactive Power Optimization Using Sensitivity Gradient Methods in Smart Grids

This paper presents a novel fault-tolerant distributed control framework for integrated active/reactive power optimization in microgrids. The architecture employs Distributed Agent Controllers (DACs) equipped with -sensitivity-based gradient methods, establishing a dual-module control structure with separate but coordinated Active Power Control (APC) and Reactive Power Control (RPC) components. The framework introduces four key advances: 1) an intelligent priority-based load-shedding mechanism within APC that maintains service quality, 2) an adaptive generation dispatch scheme coordinated with RPC, 3) decentralized optimization algorithms enabling local decision-making at each DAC, and 4) a resilient communication architecture with self-healing capabilities for continuous operation during contingencies. Implementing a multi-layer real-time approach for load management, the system dynamically evaluates consumption priorities, grid conditions, and generation constraints. Comparative simulations demonstrate superior voltage stability maintenance, reduced unnecessary load shedding, and enhanced power distribution efficiency versus conventional methods. The solution’s distinctive advantages include: 1) Grid resilience through reduced central communication dependence, 2) operational adaptability across diverse scenarios, and 3) scalable architecture for modern microgrid applications.