Reference Tracking using a Non-Cooperative Distributed Model Predictive Control Algorithm

Abstract

In this paper, a non-cooperative distributed model predictive control (DMPC) algorithm for tracking constant references is developed and evaluated. As such, an augmented model is employed (i.e. the control loop is embedded with integrators) and the augmented state contains the state increments and the error between the reference and the predicted output. The algorithm is tested in real life experiments on the quadruple tank process with non-minimum phase behaviour. The experimental results show acceptable performance index for the DMPC method when compared with the centralized approach.