Title: Design of a self-tuning double EWMA controller for MIMO processes

Authors: Ming-Shan Lu; Shi Jer Wang

Addresses: Department of Industrial Engineering and Management, National Yunlin University of Science and Technology, 123 University Road, Section 3, Douliu, Yunlin, Taiwan ' Department of Industrial Engineering and Management, National Yunlin University of Science and Technology, 123 University Road, Section 3, Douliu, Yunlin, Taiwan

Abstract: The design of time-varying multivariate discount factors for the double EWMA controller is essential for ensuring the consistency of output qualities. In this research, a self-tuning double EWMA controller for MIMO processes is proposed. First, the process estimator is used to obtain the online process parameters at each run based on the recursive least square method. Then, based on the estimated process model, the double EWMA controller is developed to generate the control inputs for compensating the variations of the process at each run. A genetic algorithm searches for the approximately optimal discount factors within the stable region at each run and the objective is to minimise the cost of the outputs deviating from targets and the adjustment amounts of the control inputs between each run. The obtained discount factors are automatically adjusted to adapt to the process variations and disturbances and to maintain the process running at controlled conditions.

Keywords: double EWMA control; exponentially weighted moving average; controller design; self-tuning control; discount factors; genetic algorithms; MIMO processes; recursive least squares; parameter estimation; process variations.

DOI: 10.1504/IJISE.2015.072727

International Journal of Industrial and Systems Engineering, 2015 Vol.21 No.4, pp.438 - 457

Received: 08 Oct 2013
Accepted: 14 Mar 2014
Published online: 28 Oct 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article