An adaptive distributed architecture for multi-agent state estimation and control of complex process systems

Abstract

A multi-agent integrated distributed moving horizon estimation (DMHE) and model predictive control (DMPC) framework is developed for complex process networks. This framework utilizes an adaptive spectral community detection-based decomposition approach for a weighted graph representation of the state space model of the system to identify the optimal communities for distributed estimation and control. As the operating conditions of the process network change, the system decomposition adjusts, and the estimation and control agents are reassigned accordingly. These adjustments enable optimizing the integrated DMHE and DMPC architecture, enhancing robustness and closed-loop system performance. The effectiveness of the proposed adaptive distributed multi-agent estimation and control framework is demonstrated through a benchmark benzene alkylation process under various operating conditions. Simulation results show that the proposed multi-agent approach enhances closed-loop performance and computational efficiency compared to traditional system decomposition methods using unweighted hierarchical community detection.