A Dual-Window Distributed Proportional-Integral Control for Large-Scale Discrete-Time Systems

Abstract

Large-scale discrete-time systems arise in many practical settings, including power grids, multi-agent robotic networks, and industrial process control. Centralized control is impractical due to high communication overhead and computational complexity, motivating distributed proportional-integral control strategies. However, existing designs apply a uniform integral window to both local and neighboring terms, failing to reflect the physically distinct sampling and communication characteristics of large-scale networks. To address this, a dual-window distributed proportional-integral controller is proposed, assigning independent window depths to the local and neighboring integral actions to accommodate heterogeneous network characteristics without additional communication overhead. Non-increasing weight sequences are incorporated to prioritize recent state information, improving transient response while maintaining analytical tractability. Stability and convergence are established via a composite Lyapunov function, yielding decentralized linear matrix inequality conditions. Simulation results validate the effectiveness of the proposed controller and demonstrate its superiority over existing control strategies across multiple network topologies and window configurations.