Gipronickel Institute LLC, St. Petersburg, Russia

D. V. Lyalinov, Senior Researcher, Geometallurgical Laboratory, e-mail: LyalinovDV@nornik.ru

Nornickel Technical Services, St. Petersburg, Russia.
E. O. Trehsvyatskaya, Leading Geologist, e-mail: TrehsvyatskayaEO@nornik.ru

The article presents experience in developing a methodology for generating dynamically updated geometallurgical models incorporating the predictive ore processing data at the Zhdanov deposit (Pechenga ore field). The relevance of the work is due to the need to obtain predictive data to stabilize the operation of the concentrating plant. The recoverable contents of nickel, copper and cobalt were used as the predictive indicator, since these indicators are additive. An indirect calculation method was used, based on the drill hole assay data of detailed and mining exploration. The beneficiation indicators were predicted using machine learning ensemble methods. A set of geostatistical studies was performed. Based on the obtained results, statistically substantiated parameters of the optimal network of technological testing were proposed. A script was developed in the Micromine environment for updating block models when data appears on new exploratory wells and / or performed laboratory technological studies. A comparison of the predictive indicators of block models with the results of daily operation of the concentrating plant for two months was carried out.

1. Kravtsova O. A., Maksimov V. I., Lebedeva A. A., Koptev K. V. The practice of geological mapping of the Zhdanov deposit ores in the Kola peninsula. Tsvetnye Metally. 2020. No. 12. pp. 39–44.
2. Mishulovich P. M., Petrov S. V. Methodological aspects of generating geometallurgical models of mineral deposits. Vestnik SPbGU. Nauki o Zemle. 2019. Vol. 64. Iss. 2. pp. 249–266.
3. Maltsev E. N. Geological and technological modeling of physical and mechanical properties of ores and indicators of solid mineral beneficiation. Nedropolzovanie XXI vek. 2020. No. 6. pp. 96–103.
4. Kozlova M. A., Ryabtsev D. A. Modern approach to geometallurgical mapping of ore deposits. Geologiya mestorozhdeniy poleznykh iskopaemykh. 2017. No. 1. pp. 23–30.
5. Kalashnikov A. O., Ivanyuk G. Yu. Prediction of the composition of ore mine rals based on the chemical composition of ore using artificial neural networks (using the Kovdor baddeleyite-apatite-magnetite deposit as an example). Gorny informatsionno-analiticheskiy byulleten. 2019. No. S37. pp. 485–492.
6. Both C., Dimitrakopoulos R. Applied machine learning for geometallurgical throughput prediction — A case study using production data at the Tropicana Gold mining complex. Minerals. 2021. Vol. 11. 1257. DOI: 10.3390/min11111257
7. Ortiz J. M., Cevik S. I., Avalos S., Kracht W. et al. Machine learning and deep learning in predictive geometallurgical modeling. PDAC. Toronto, ON, March 4, 2020.
8. Both C. et al. Geometallurgical prediction models of processing plant indicators for stochastic mine production scheduling. IFAC PapersOnLine. 2022. Vol. 55, Iss. 21. pp. 162–167. DOI: 10.1016/j.ifacol.2022.09.261
9. Carrasco P., Chiles J.-P., Seguret S. Additivity, metallurgical recovery, and grade. 8th International geostatistics congress. Santiago, Chile. 2008.
10. Breiman L. Random Forests. Machine Learning. 2001. Vol. 45. pp. 5–32.
11. Friedman J. H. Greedy function approximation: A gradient boosting machine. Annals of Statistics. 2001. Vol. 29. pp. 1189–1232.
12. Lechuti-Tlhalerwa R., Coward S., Field M. Embracing step-changes in geoscientific information for effective implementation of geometallurgy. Journal of the Southern African Institute of Mining and Metallurgy. 2019. Vol. 119.