Review of machine learning application in mine blasting

Mine blasting has adopted machine learning (ML) into its practices with the aims of performance optimization, better decision-making process, and work safety. This study is aimed at reviewing the status of ML method applications to mine blasting issues. One of the most important observations of this research highlights the developed ML methods such as hybrids/ensembles, outperforming the other methods at 61% of the sample of case studies. The first section provides a background on the application of ML methods in mining. Two sections of the review provide the trends in the application of ML methods and the utilization of input parameters in surface and underground blasting problems. The appraisal reveals an increase of hybrid/ensemble or highly developed ML methods for the top four blast issues on the surface (72%) and underground (45%). The sample of studies reviewed indicated through graphical/statistical means a continuing increase in hybrids/ensembles’ use mirrored by high research output for the top four surface blast issues. This is contrasted by a low rate of research in underground blasting, under the encountered operational conditions applied. Regarding the input parameters, controllable parameters (blast design and geometry) were recognized to be steadily used in surface blasting for the top four surface blast issues, along with less involvement from the uncontrollable parameters (geological and geotechnical parameters). On the contrary, underground blasting has a slight involvement of the uncontrollable parameters more than the controllable parameters, in the top four blast issues. In the final section of this paper, the review offers a discussion of the current state of research under the encountered limits and where the efforts should be focused concerning ML methods applied, input parameters involved, and the challenges faced. Such high levels of performances are in demand in a highly complex mining environment. Persistent research, and development of mining employees’ technological skills alongside an increased awareness among mining industry of the benefits of highly developed ML techniques, is greatly needed at this stage. This would establish the role of highly developed ML methods in improving both the blasting process and the overall decision-making and mining management.