A multiagent approach using model-based predictive control for an irrigation canal

Abstract

This paper presents the application of themultiagent paradigm to a distributed model-based predictive control (DMPC) scheme in order to improve its fault tolerance, give it the ability to dynamically adapt its strategy to optimize energy consumption, and to allow it to scale up. This approach is illustrated in the control of a canal simulated using realistic, physics-based 1Dmodels inMatLab. The individual agent behavior, based onDMPC, and themultiagent composition mechanism are described. Presented simulation results illustrate the ability of the proposed control scheme to adapt to a hardware failure and to take global strategies into account.