Statistical analysis of modified predictive control of non-minimum phase system

Abstract

Testing statistical hypothesis in comparison of corresponding pairs of signals in control has not been widely used both in practice and research. In this paper, pairs of signals obtained during control before and after a modification of a control algorithm are paired tested by quantitative methods. In previous research, the authors used this type of mathematical analysis with regards to the control quality in an analysis of signals in multivariable MPC with a modified optimization strategy. The control quality was slightly influenced in favour of the decreasing the computational complexity of the MPC control algorithm. In this paper, results of a predictive control of a non-minimum phase system with an elimination of the undershoot are analysed in detail. The control algorithm was modified by a particular setting of control constraints. The aim of this modification was suppression of the undershoot. The quality of control with the original and modified control algorithms is analysed by testing hypothesis. Particular signals are compared using the testing hypotheses on the statistically significant differences. © 2019, World Scientific and Engineering Academy and Society. All rights reserved.