Technical Notes - Intelligent decision-support system for mine managers

Mine managers rely on their experience, section foreman's daily reports and occasional time studies to make important production, maintenance and forecasting decisions. This information is often subjective and imprecise, consumes expensive engineers time, is often biased and is restricted to short time periods. Managers need better information. Furthermore, if a recently purchased machine is being evaluated, a large increase (>10%) in productivity may be obvious, but small increases are usually masked by other variables. In today's highly competitive markets, cumulative small changes are important, especially since managers have already exhausted most available, obvious avenues of improvement. A technique to separate these charges from effects of other variables and to quantify the target effect is necessary. CSM researchers are addressing this need by developing an intelligent decision support system (IDSS). It provides key production and maintenance information to support management decisions by recognizing patterns in mining machine sensor data and representing knowledge about mine management, operations, layout and equipment. A continuous miner section at Western Fuels-Utah Inc.'s Deserado Mine in Rangely, CO is the test facility for the research. Power transducers located in the sections load center provide kilowatt versus time data from continuous section equipment. The IDSS pattern recognition algorithms classify the kilowatt versus time patterns into individual machine operating modes. It produces an output similar to a typical industrial engineering time study but continuously, automatically and without an engineer in the section. It is relatively easy for a time-study engineer to identify the stop and end points of a cycle element and classify it by looking at the machine and glancing at a stop watch. The problem is that monitoring system hardware and software do not have these unique capabilities. Consequently, IDSS algorithms must automatically define the begin and end points of machine tasks and identify the task by recognizing patterns in kilowatt versus time data. However, many mines have file drawers of time-study type data that are not very useful. So, IDSS contains the knowledge representation subsystem (IDSS/KRS) to produce short reports containing only key information. This is difficult because mine environments continually change and information requirements differ for each management level. Consequently, IDSS/KRS must learn and adapt to new conditions in a dynamic mining environment. Western Fuels-Utah owns and operates the Deserado Mine, located about 19 km (12 miles) northeast of Rangely, CO (Upadhyay, 1988; King and Eros, 1988). One longwall and two continuous miner sections produce about 900 kt (1 million st) of coal annually for the Deseret Generation and Transmission Cooperative's 4000 MW power plant located near Bonanza, UT. Three mine monitoring systems control the underground mine, the preparation plant and the unit train load out. Pattern recognition IDSS/PRS produces a listing of begin and end times for each machine operating event as shown is Table 1.