Mining Ventilation: Expert System Based Operative Control

Problems of how to use artificial intelligence (AI) methods and expert systems (ES) to control complex industrial projects have lately become significant in practical terms. Mining ventilation systems undoubtedly belong to complex multi dimensional objects of control with stochastic nonstationary and nonlinear processes. Formal models based on automated ventilation control, do require awkward algorithmic solutions, moreover, a number of tasks become the operator's headache, and the quality of their solution depends on his experience and intuition. These tasks include: - identifying pre-accidental situations and determining their causes; - making well-grounded decisions with regard to changing ventilation regimes, forecast of mining operations taken account of; - controlling ventilation systems in emergency situations, e.g. when in accident, etc. It is in this context that we have for a number of years been investigating how to practically use AI methods for operative control in dynamic aero-and-gas mining processes, one of our research directions being devoted to working out a consulting programme for the ventilation system operator. Our researching and modeling problem was identification of aero-and-gas situation in different parts of ventilation network and its interpretation in order to determine possible causes for gas regime changing and unexpected consequences. We would single out the following stages in this task: I. Operative information procession coming to the data base from the sensors and analysis of hypothetical versions (using the associative rules base) with regard to factors which have caused significant parameter fluctuations (e.g. C%(t), H(t), Q(t) etc.)., 2. Selection of hypotheses in the knowledge base and forming conclusions in accordance with reliability scale. 3. Clarifying the obtained conclusions in dialogue with the operator. 4. Correcting the associative rules base.