Fuzzy Clustering's Research In Prediction Of Coal And Gas Outburst Accident

The subject of this paper is that using the fuzzy c-means(FCM) clustering algorithm of fuzzy clustering to analyze a large amount of historical data, excavate the key indicators of coal and gas outburst and establish a prediction system of coal and gas outburst to provide theoretical support for the prediction of coal mine accidents and protection for the safety of the miners. Fuzzy c-means clustering algorithm is the most widely used algorithm, for fuzzy c-means algorithm, the over-reliance on the choice of initial cluster centers, using subtractive clustering to determine the initial cluster center .for the original fuzzy c-means algorithm need to determine the number of categories of data sets in advance that damage its non-supervisory, using subtractive clustering to determine the maximum number of clustering, and then combined with a new validity function to determine the optimal clustering number -Copt, so it needn't initialize repeatedly, thereby enhancing the efficiency of clustering. Adopt the method which is weighting sample data's features to distinguish the each index sample's separate contribution to classification.Using the improved fuzzy c-means algorithm to establish the coal and gas outburst's prediction system, the prediction result show that the model has good practical value.