An Expert System Philosophy Used In The Early Warning Of Large Goaf Falls In A Coal Mine

A longwall-based coal mine in Australia, referred to in this paper as AusMine Colliery, experienced windblasts resulting from goaf falls. Windblasts were the cause of serious injuries and threatened the continued operation of the mine. A microseismic early warning system was installed and a significant success rate in providing early warning of large roof instability resulting in high velocity windblasts, has been achieved. Early warning of goafing is accepted as an operational tool on the mine. Severe demands are placed on the operator in terms of 'real time' interpretation. The development of an expert system allowed for a probabilistic approach to achieve automatic alarms. This paper will quantify the success, failure and false alarm rates achieved by manual operation and automatic expert system-based decisions making, respectively.