MPC automation system; Model predictive control; Parametrization; Horizons setting; Control quality; Statistical analysis; Statistical significance

In automation systems, a parametrization of a controller in a part of its synthesis has an important influence on an accordance to an improving the control quality aspects and on a decreasing the computational complexity. In the field of the process control, as one of the modern control methods, the Model Predictive Control (MPC) has been considered. The MPC strategy is one of a novel and modern approach. In the MPC, the control process has been influenced by parameters dependent on the strategy of the receding horizons. These horizons can be set by programmers of the controller. However, the parametrization has not been so widely bound on the statistical analysis of the MPC progresses of the control quality criterions yet. The quantitative research techniques can be one of the appropriate approaches of decision, which parameters are suitable. In this paper, the statistical methods of the testing differences between the MPC control criterions are proposed as an extended method with regards to the guarantee of a statistical significance. Particularly, the testing differences of the criterions of the MPC automation system is demonstrated on the control of the multivariable model of the process including a consideration of the parametrical (ANOVA) or non-parametrical (Kruskal-Wallis) statistical approaches based on the significance level 0.001.