Pattern classification of continuous mining duty cycle data for an intelligent decision support system

The microcomputer based monitoring and control system at Western Fuels Utah Incorporated Deserado Mine is collecting data on production equipment power usage that can input to an intelligent production and maintenance management decision support system. The problem is to develop techniques that will classify patterns of kilowatt versus time data in the variable mine environment. This paper will focus on mathematical algorithms developed to date for production and maintenance information from a continuous miner section. These algorithms classify the kilowatt vs. time patterns into machine operating tasks. Data convolution with Fourier, least squares, running average, and other functions have been tested to determine which if any enhances classification. Simple and complex single and multiple step classification techniques based on parsing algorithms and statistical modeling have also been studied. However, because the research is ongoing, more optimal pattern classification techniques may be found in the future.