Distributed Automatic Generation Control subject to Ramp-Rate-Limits: Anytime Feasibility and Uniform Network-Connectivity

2025-09-08

AI Summary

Existing AGC optimization methods often neglect generator ramp-rate limits (RRLs), leading to infeasible solutions or failure to converge to the optimum under practical operational constraints. To address this, we propose a distributed automatic generation control (AGC) method that rigorously enforces RRLs. Leveraging a multi-agent system and an information-sharing network, our approach employs a consensus-based distributed optimization algorithm augmented with a sign-function-based acceleration mechanism, ensuring strict satisfaction of power balance, generation limits, and ramp-rate constraints at every iteration. To the best of our knowledge, this is the first distributed AGC framework achieving both *anytime feasibility*—i.e., physical feasibility upon termination at any iteration—and explicit RRL modeling, while remaining robust to intermittent communication link failures. Theoretical analysis and numerical simulations demonstrate its fast convergence and stable tracking performance under dynamic communication topologies.

Technology Category

• Search and Optimization: Distributed Search
• Multiagent Systems: Agent Communication
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

This paper considers automatic generation control over an information-sharing network of communicating generators as a multi-agent system. The optimization solution is distributed among the agents based on information consensus algorithms, while addressing the generators' ramp-rate-limits (RRL). This is typically ignored in the existing linear/nonlinear optimization solutions but they exist in real-time power generation scenarios. Without addressing the RRL, the generators cannot follow the assigned rate of generating power by the optimization algorithm; therefore, the existing solutions may not necessarily converge to the exact optimal cost or may lose feasibility in practice. The proposed solution in this work addresses the ramp-rate-limit constraint along with the box constraint (limits on the generated powers) and the coupling-constraint (generation-demand balance) at all iteration times of the algorithm. The latter is referred to as the anytime feasibility and implies that at every termination point of the algorithm, the balance between the demand and generated power holds. To improve the convergence rate of the algorithm we further consider internal signum-based nonlinearity. We also show that our solution can tolerate communication link removal. This follows from the uniform-connectivity assumption on the communication network.

Problem

• Addressing generators' ramp-rate-limits in distributed optimization
• Ensuring anytime feasibility with power balance at iterations
• Maintaining solution convergence under network connectivity changes

Innovation

• Distributed multi-agent optimization with consensus algorithms
• Addressing ramp-rate-limits and box constraints for feasibility
• Uniform network connectivity enabling communication link tolerance