National University of Science and Technology “MISIS” (Moscow, Russia)

N. V. Osipova, Cand. Eng., Associate Prof., Dept. of Information and Communication Technologies of the Institute of Computer Sciences, e-mail: nvo86@mail.ru

The article describes the main problems that arise in the management of the wet magnetic separation process used in iron ore enrichment. It is shown that the content of magnetite iron in the input stream, pulp density and average particle size have the main influence on it. The pulp containing crushed ferruginous quartzite from the deposits of the Kursk magnetic anomaly was selected as the object of research. To recognize the technological properties of the pulp entering the wet magnetic separation process, it is proposed to use cluster analysis based on the Kohonen artificial neural network, which consists in dividing collections into groups which are homogeneous according to specified characteristics: the content of magnetite iron in the input stream, pulp density and average particle size. Each group corresponds to a specific cluster that characterizes technological properties. It combines the values of the indicators and each of them is within certain limits. To test the performance of the clustering algorithm, collections were generated in the Matlab application software package with the data on the content of magnetite iron, pulp density and average particle size, representing random values with a normal distribution law, the parameters of which were calculated based on reference information and set different for each of the three groups of properties. The training of the Kohonen artificial neural network was carried out in the Matlab application Neural Toolbox. Its testing took place in the “sliding” window mode. The simulation results showed that the clustering process is conducted qualitatively, because the coefficient of cofenetic correlation calculated in the “sliding” window mode has sufficiently high values exceeding 0.8.

1. Osipova N. V. Model for optimal control of a magnetic separator based on the Bellman dynamic programming method. Chernye Metally. 2020. No. 7. pp. 9–13.
2. Osipova N. V. Investigation of the possibility of obtaining concentrate production targets based on a mathematical model of an ferrum ore processing site. CIS Iron and Steel Review. 2023. Vol. 25. No. 1. pp. 4–9.
3. Osipova N. V. Automatic control system for wet magnetic separation of iron ore. Gornyi Zhurnal. 2019. No. 1. pp. 62–65.
4. Bogdanov O. S., Nenarokomov Ju. F. Handbook for mineral processing. Concentrating plants (Vol. 2). Moscow: Izdatelstvo “Nedra”, 1984, 360 p.
5. Absatarov S. H., Moseikin S. S. Features of the physico-mechanical properties of the mineral varieties of ferruginous quartzites of the Lebedinsky deposit. Gorny informatsionno-analiticheskiy byulleten. 2008. No 2. pp. 285–291.

6. Aleksandrova T. N., Chanturiya A. V., Kuznetsov V. V. Mineralogical and technological features and patterns of selective disintegration of ferruginous quartzites of the Mikhailovskoye deposit. Zapiski Gornogo instituta. 2022. Vol. 256. pp. 517–526. DOI: 10.31897/PMI.2022.58
7. Ismagilov R. I., Baskaev P. M., Ignatova T. V., Shelepov E. V. The prospects for expanding the iron ore mineral and raw material base through the processing of oxidized ferruginous quartzite of the Mikhailovskoe deposit. Obogashchenie Rud. 2020. No 3. pp. 19–24.
8. Zhang Y.-S., Sun X.-T., Yuan H. Yang G. Determination of potassium, sodium and zinc in iron ore by X-ray fluorescence spectrometry with pressed powder pellet. Yejin Fenxi/Metallurgical Analysis. 2019. No. 39 (1). pp. 72–76. DOI: 10.13228/j.boyuan.issn1000-7571. 010378
9. Topchaev V. P., Topchaev A. V., Lapidus M. V. Industrial in-line automatic granulometer PIK-074P as the basis of automatic control and quality materials grinding control system. Tsvetnye Metally. 2015. No. 9. pp. 48–52.
10. Pelevin A. E. Improvement of concentrate quality by using fine screening in iron ore grinding stage. Chernye Metally. 2023. No. 10. pp. 4–9.
11. Timashev V. P. Formation of stable quality of iron ore concentrate. Dissertation … Cand. Eng. Moscow, Moscow Mining Institute, Scientific Research Institute on the Problems of the Kursk Magnetic Anomaly. 1984. 173 p.

12. Singh D. Cluster Space Among Labor Productivity, Urbanization, and Agglomeration of Industries in Hungary.
Journal of the Knowledge Economy. 2022. No. 13 (2). pp. 1008–1027. DOI: 10.1007/s13132-021-00726-9
13. Zhou G., Miao F., Tang Z., Zhou Y., Luo Q. Kohonen neural network and symbiotic-organism search algorithm for intrusion detection of network viruses. Frontiers in Computational Neuroscience. 2023. No 17. pp. 1–15.
14. Oti E. U., Olusola M. O. Overview of agglomerative hierarchical clustering methods. British Journal of Computer, Networking and Information Technology. 2024. Vol. 7. Iss. 2. pp. 14–23.
15. Rankovic N., Rankovic D., Lukic I., Savic N., Jovanovic V. Unveiling the Comorbidities of Chronic Diseases in Serbia Using ML Algorithms and Kohonen Self-Organizing Maps for Personalized Healthcare Frameworks. Journal of Personalized Medicine. 2023. Vol. 13 (7). 1032. pp. 1–19. DOI: 10.3390/jpm13071032
16. Latypova D. S., Tumakov D. N. Clustering of handwritten digits by Kohonen neural network. Programmnye sistemy: teoriya i prilozheniya. 2022, Vol. 13. No. 3 (54). pp. 225–239. DOI: 10.25209/2079-3316-2022-13-3-225-239
17. Uvaisov S. U., Chernoverskaya V. V., Dao An Kuan, Nguen Van Tuan. Kohonen’s algorithm in problems of classification of defects in printed circuit assemblies. Russian Technological Journal. 2021. Vol. 9. No. 4. pp. 98–112. DOI: 10.32362/2500-316X-2021-9-4-98-112