Implementation of a Distributed Optimal Predictive Control in a Quadruple Tank System

Abstract

This paper presents the development of a distributed optimal model predictive control (DOMPC) applied to a quadruple tank system. The DOMPC is based in an optimal model predictive control (OMPC) approach. In order to distribute the control laws, the DOMPC divide the controller in s local controllers. The local controllers are communicated by a local area
network (LAN). For the control input calculation, it is assumed that this LAN introduces a delay of one sample. A Labview virtual instrument, that allows to implement and configure the DOMPC controller, is developed in order to apply the controller in a quadruple tank process. This virtual instrument uses an intuitive graphical interface to adjust the controller parameters and incorporates a simple but effective strategy to cope with information losses in the LAN. The development is a distributed scheme that has the benefit of coordinate the tasks from local controllers to achieve the global control objective. Finally, the results obtained are compared with a centrilized OMPC in order to evaluate its performance and its applicability.