Innovative dynamic short-term, medium-term and long-term mine planning strategies incorporating new automation and data analytics technologies

点击翻译按钮获取中文摘要

Mining industry faces unique challenges where the metal prices fluctuate by almost 400% and the grade of ore declines over the years. The orebody samples collected in the exploration stage are sparse and prone to random errors. Also, the financial parameters such as metal price, discount rate and costs are always changing and difficult to predict. Consequently, long-term and medium-term operational decisions are made in the absence of sufficient information. This brings high levels of uncertainty, which leads to unexpected consequences, creating a need for flexible long-term plans that are able to change, based on future circumstances and prescient medium-term plans that are designed for all possibilities. On the other hand, with the inflow of big data from new sources, such as automated equipment and the industrial internet of things (IIoT) technology, there is an opportunity to make short-term plans with higher precision. However, the abundance of information from many sources and extensive usage of autonomous equipment pose new challenges in mining such as anomaly detection, integrating of this information into short-term mine planning and considering extreme conditions. The long term plan of the proposed research plan is to adapt the progress in machine learning and optimization and to capitalize on IIoT sensors of mining equipment to develop new dynamic, adaptable mine planning strategies that make more informed decisions in the presence of uncertainty for the long term and sensor data for the short term in order to sustain the operations and increase the profit despite ore price fluctuations and grade deprivation. This goal is addressed in the short term by focusing on two issues: (1) Establishing planning approaches that understand and handle uncertainty: Identify the sources of uncertainty and their combined effects on the mine plan. Using this in-depth understanding, define new approaches to long-term mine planning accounting for different scenarios of grade distribution of the orebody, costs and commodity prices. (2) Solving potential problems of mine automation and integrating it into mine planning: Use data analytics to validate, integrate and analyze the data coming from the different sensors and use this information in short-term mine planning with state-of-the-art machine learning techniques. Account for communication and sensor failures by developing methodologies to handle misreadings, miscommunications and interruptions. In the proposed research plan, 2 PhD, 3 Master's and 2 Undergraduate students will be trained. Furthermore, novel strategies for better adaptation to uncertainty and mine automation will be developed, which are expected to have a significant impact for the mining industry in Canada and abroad.