Research On Fuzzy Clustering Analysis To Prediction Parameters Of Mining Subsidence

With the increase of the "three-below" coal mining intensity, geological hazards due to mining subsidence are increasingly serious. To reduce mining subsidence disaster caused by the "three-below" mining, whether it can accurately make a prediction of surface deformation is absolutely vital, and how to select prediction parameters becomes a crucial issue. Aming at unscientific and uncertainty problems in the existing methods of obtaining prediction parameters, the paper determines the complicated relationship between prediction parameters and the geological and mineral features by applying fuzzy clustering analysis with regression analysis method. The paper not only adopts combination method of feature selection to determine the main geological and mineral features of mining subsidence, but also proposes an improved fuzzy clustering algorithm, through which sample stations of mining subsidence are divided into four similar phenomenon groups. From the perspective of fuzziness in membership relation, it establishes a segmentation model based on the judgement algorithm of membership relation, through which the paper overcomes the shortcomings of the existing methods. Taking Xuzhou mining area and Northwest Liaoning mining area as the analysis examples, the point is proved accurate and reliable that the segmentation model is suitable for the single and regional mining area, which provides a method of selecting prediction parameters for a new mining area without a priori information. The results show that the model is of great significance for theoretical value and guiding effect in the perspective of prediction parameters, it also provides a new technology method and strong information support for accurate prediction of mining subsidence, disaster warning and reasonable design of "three-below" mining in the whole country's mining areas.