Predictive control of processes with utilization of artificial intelligence elements

Predictive process control is a method of regulation suitable for controlling various types of systems, which is based on the idea of using the prediction of future system behavior and its optimization. Normally, a system model is used to predict behavior, and therefore it is necessary for the correct function of predictive control to make its correct selection and determine its parameters so that the controlled system is described as accurately as possible. Another advantage of predictive control is the possibility of including signal restrictions directly in the controller. The result is the application of some elements of artificial intelligence in suitable areas of predictive control, especially the use of simple evolutionary algorithms in optimization and neural networks as nonlinear models. One of the chapters of the article describes the possibilities of using these elements. It is proved that in addition to classical optimization algorithms, it is also possible to use simple evolutionary algorithms for optimization of prediction, while the computational complexity can be comparable depending on the type of solved problem and settings. The article describes a suitable selection of model systems with slow dynamics, their derivation, and the creation of nonlinear models in the form of scalable neural networks. The potential advantage of this approach for the control of systems that are difficult to describe or for the control of systems whose mathematical-physical description is not known. The chapter of the article also deals with the possibility of using the found models on real systems and determining the necessary conditions and requirements for their application. © 2020 Danube Adria Association for Automation and Manufacturing, DAAAM. All rights reserved.IGA/CebiaTech/2020/003, IGA/FAI/2020/003; Ministerstvo Školství, Mládeže a Tělovýchovy, MŠMT: LO1303, MSMT-7778/2014; European Regional Development Fund, ERDF: CEBIA-Tech ED2.1.00/03.008