Synthesizing models of nonlinear dynamic objects in c oncentration on the basis of Volterra-Laguerre structures

Abstract

Purpose. Enhancing energy efficiency and quality of automated control of the technological concentration line, increasing extraction of the useful component into concentrate while processing iron-bearing ores of various mineralogical and technological types through developing principles and approaches to distributed optimal control over interrelated processes in concentration production on the basis of the dynamic space-time model. Methodology. Based on the assumption that final results of concentration plant operation depend on a set of input parameters and results of functioning of interrelated nonlinear dynamic objects, the authors suggest an improved approach to simulating concentration processes for iron ore materials on the basis of Volterra-Laguerre structures by using input signals of certain technological stages characterizing granulometric composition of processed ore. Findings. It is found that while synthesizing models of nonlinear dynamic objects of concentration, it is expedient to apply Volterra structures with the simulation error not exceeding 0.039 under the mean square deviation of 0.0594. Volterra models projected onto orthonormal basis functions enable simplifying parameterization and reducing sensitivity of models to noises. Among other orthonormal functions, Laguerre functions are reasonable to use. All this allows minimizing the number of model parameters in the course of their identification. Originality. The method of identifying nonlinear dynamic objects of concentration on the basis of the space-time Volterra model is improved. This model is different from available ones by its projection onto orthonormal Laguerre basis functions to increase its robustness to noises. Practical value. Testing results enable deducing efficiency of the space-time Volterra model in the condition space by means of the Laguerre network, thus increasing accuracy of simulation under noises as compared to the Volterra model through reducing the simulation error by 18.11 % under 40 iterations of identification. The experimental check of identification accuracy by means of the Volterra-Laguerre model in the iron content control system in various points of the technological concentration line confirms efficiency of the given method.