Senko A. Information technology decision-making process grinding on the basis of indirect determination of the strength of ore

In the thesis, the actual scientific and technical problem of developing information technology for the inverse prediction of the ore strength parameter in the enrichment section operation was solved using a combination of clustering methods and a predictive neural network. The general problems of process management in mining ore factories are covered, existing methods of determination of parameters of input raw materials are considered. Their advantages and disadvantages are revealed. The approach is based on the reverse forecasting by processing the accumulated statistical data. The results of researches of dependencies between static and dynamic characteristics of wet magnetic enrichment complexes, as well as statistical characteristics of perturbations on the process of magnetic separation, are summarized and presented. The expediency of using a set of parameters is suggested, which increases the correlation and allows to speak about the existing patterns. An algorithm for forming a training sample based on methods of cluster analysis of Microsoft Clustering algorithm is developed. An algorithm for forecasting the input parameters of the enrichment section with the use of a three-layer neural network with counter-recognition without feedback is developed. An information technology of a decision support system based on a combination of clustering methods and the use of a predictive neural network has been proposed, which allows a specialist to promptly receive recommendations on making decisions regarding the behavior of an object. The results were implemented at Kryvyi Rih Industrial Investment Company, Kryvyi Rih Institute of Automatics.