USE OF NEURAL REGULATORS IN CONTROL SIMULATION OF CRUSHING STAGE IN CONDITIONS OF MINING AND PROCESSING PLANT

Abstract

Abstract. The article is devoted to possibility of using standard types of neuroregulators that offer the MATLAB & Simulink environment in modeling control of technological process, namely grinding stage, by applying consistent intelligent control under uncertainty. Use of artificial intelligence technologies in mining is quite relevant at this time. Unlike "classical" deterministic control systems based on rigid algorithms use (or clear logic), systems using artificial intelligence have properties of learning and self-learning (that is, accumulation and generalization of experience). Use of artificial neural networks to model and identify control object is approach that is usually considered as alternative to methods based on physical or technological principles. In particular, this concerns possibility of using neural networks and fuzzy logic to control of crushing-grinding and enrichment of minerals technological processes. Paper considers three possible types of controllers offered by MATLAB & Simulink, namely the NN Predictive Controller, the NARMA-L2 autoregressive control controller and the Model Reference Controller controller. Each of the considered regulators can be applied at modeling of technological process, but expediency of use of this or that type, first of all depends on character of technological process. During the simulation, the possibility of controlling the technological process with the help of artificial intelligence (regulators based on neural networks) was investigated. Analysis of results of modeling three types of neuroregulators showed that the most appropriate for modeling the control of the technological process of grinding is the use of a regulator type NARMA-L2.