Development of Sensors for Advanced Mining Systems

点击翻译按钮获取中文摘要

The research program is aimed at developing advanced mining systems through the application of sensors. Conventional mining is static, plan based, subjective and time consuming. In conventional mining, the ore resources is defined based on relatively small amounts of data resulting from the analysis of drill core during the early stages of resource development. Geo-statistics are then applied to create the ore body model and mine plan for the life to the operation. Studies have shown that in conventional mining systems valuable ore is often sent to the waste dump and waste misreports to the process plant; one case study on a large open pit copper mine indicated that 30% of the rock in the waste dump was of ore quality grade. In advanced mining systems, sensors would mounted on material handling equipment such as shovels, conveyors and stockpiles where they would collect data that would be converted into meaningful information using calibrated algorithms for real time decision making. Research has shown that sensors that characterize ore as it is being handled can improve operation activities such as production scheduling, grade control and process plant control settings making them more flexible and objective orientated. Sensors allow simultaneous decision making in real time thereby creating an integrated dynamic mining and processing system.The enabling technology for the advanced mining systems are the sensor systems that can characterize important properties of the ore. The research is aimed at advancing the sensor systems to quantify properties of the ore that affect operational goals such as throughput, metal production and resource utilization. Ore properties of interest include lithology/ ore type, hardness, fragmentation, grade and the presence of impurities. Understanding the ore heterogeneity is an important focus of the research because it indicates the amenability of the ore to a sensor based mining system; without heterogeneity of properties such as grade, hardness or lithology, there is no opportunity for classification. The challenge is to develop algorithms that can accurately relate the sensor response to the ore property. A range of sensors will be used including x-ray fluorescence, x-ray transmission, laser induced breakdown spectroscopy and electromagnetic. The sensor responses will be related to ore properties through the development of algorithms that can be generated using statistical modelling (multivariable regression) or Artificial Intelligence (Genetic Algorithms, Neural Network) approaches. Robust algorithms should be grounded in the relationship between the sensor response and the physical and chemical properties of the minerals in the ore.