Development of a knowledge-based system for assessing unplanned dilution in open stopes

A reliable estimation of unplanned dilution is one of the most vital parameters in open stopping mining methods as unplanned dilution and sloughing can lead to an increase of the cost production and most importantly to safety concerns. Owing to this importance, several predictive models to estimate unplanned dilution and stope overbreak have been proposed. However, some of these existing models do not provide any physical meaning of the underlying dilution problem. To overcome this limitation, a fuzzy inference system (FIS) is proposed. In this paper, a new design tool capable of assessing unplanned dilution in a reliable manner is developed. The methodology relies on a fuzzy inference system (FIS) as many epistemic uncertainties that are associated with unplanned dilution are not systematically accounted for in existing methods. To this end, a dilution database from Ridder-Sokolny mine in Kazakhstan has been compiled. It consists of unplanned dilution cases and includes stope geometry, rock mass characteristics and stope reconciliation data. Subsequently, knowledge-based FIS was implemented. Overall, the results of the FIS show good classification performance (84% of accuracy). In comparison to the dilution graph developed based on traditional method, the FIS results show better performance. Hence, it was concluded that the proposed FIS could be an alternative tool for empirical open stope design.