Optimisation-Based Tuning of Dynamic Matrix Control Algorithm for Multiple-Input Multiple-Output Processes

摘要

This work discusses tuning of Model Predictive Control (MPC) algorithms by means of some global optimisation methods. For test purposes the Dynamic Matrix Control (DMC) algorithm applied to a Multiple-Input Multiple-Output (MIMO) process with 4 manipulated and 3 controlled variables is considered. The tuned parameters include prediction and control horizons as well as the weights of the minimised MPC cost-function. Four global optimisation methods are considered: the Particle Swarm Optimisation method, the Firefly Algorithm, the Grey Wolf Optimiser and the Jaya algorithm. They are compared in terms of the ability to find the best solution and convergence. The obtained results show that global optimisation methods can be successfully used in this type of tasks.