Mobile Underground mining equipment has complex dynamics that has limited the application of online in-situ automated fault detection techniques. This class of machinery is generally subject to variable states including changing speed and load. The absence of sensitive and reliable methods for reliability analysis in this segment precludes industry from leveraging the well-established benefits of condition monitoring including the avoidance of major stoppages in operations, the optimization of the employment of maintenance and reliability staff, just-in-time parts inventories, etc.

This work focuses on extending artificial-intelligence techniques to provide automated online in-situ fault detection of the mechanical components of mobile-mining equipment. It is an understood maxim in the artificial intelligence/pattern recognition domain that there is no one best algorithm for classification of data;​ this is particularly true when one seeks to find faults in such machinery—the challenges of monitoring a hoist are similar but not equivalent to those in monitoring a load-haul dump truck. Under this consideration, a number of techniques are advanced with varying strengths and weaknesses to address the varied nature of variable-state machinery.

Algorithms are advanced that minimize the amount of training data while ensuring that the impact from all ranges of changing variables like speed and load are incorporated into the model. The effect of these techniques ultimately enables the detection of faults at earlier points in their progression in comparison to condition monitoring that is not adapted to the challenges of this problem. Finally, a software framework that facilitates the dynamic structuring of a condition-monitoring solution for a wide array of problems is presented;​ it is designed for bandwidth limited environments like the underground mining environment and is capable of supporting important data-management frameworks like the upcoming International Rock Excavation Data Exchange Standard (IREDES).