Machine learning for open-pit mining: a systematic review

ABSTRACT

Nowadays, open-pit mining is the large-scale extraction of valuable ore materials from the surface with the use of modern mining equipment. If not operated properly, various unexpected events, such as equipment breakdown, slope collapse, hazardous gas emission and land pollution, would occur. With the rapid development of computer technology and big-data science, emerging applications of machine learning could significantly improve mining predictability, feasibility, efficiency and sustainability. However, there is still a lack of up-to-date systematic literature reviews on applications of machine learning to open-pit mining. To address this issue, this study reviews over 200 relevant papers mainly published in the last five years. In this review, we initially conduct a descriptive statistical analysis of these papers according to different phases in open-pit mining. Consequently, we classify their research findings into four main categories: exploration, exploitation, production and reclamation. In addition, each main category is further divided into some sub-categories, namely, feasibility evaluation and mine design planning in exploration; mine block sequencing in exploitation; drilling, blasting, haulage and processing in production; waste control and environmental protection in reclamation. Based on such a bi-level classification, we systematically summarise promising machine learning techniques (i.e. reinforcement learning and deep reinforcement learning) and potential research opportunities (e.g. integration of machine learning and simulation for mining equipment scheduling) in real-world implementations for the mining industry.

KEYWORDS:

• Machine learning
• open-pit mining
• mine block sequencing
• drilling
• blasting
• haulage

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the National Natural Science Foundation of China [Grant No. 71871064].