Method for identifying mine gas emission abnormity

The invention discloses a method for identifying mine gas emission abnormity. Monitoring data sequences of monitoring points at different positions in a pit serve as random variables to establish a variable set, and a belief network structure is established; each variable is evenly dispersed to form a value taking interval; a least square method is adopted to finish network learning, and network parameters are calculated to quantitatively represent the dependency degree of gas flow at each monitoring point position; the joint probability and the posterior probability maximum of a monitoring data sequence of each monitoring point on different value taking intervals are calculated, the integral whole gas emission level is judged, finally the maximum posterior probability of taken gas monitoring data values is obtained through probabilistic reasoning, and mine gas emission abnormity conditions are judged by combination with historical monitoring data statistical characteristic analysis. Calculation results of th