Study On Regularity And Prediction Method Of Surface Subsidence Due To Deep Mining

As the coal resources exhausted in shallow, mining depth extends to deepgradually, deep mining subsidence and prediction become focus of mining subsidenceresearcher, research on this, have important theoretical and practical significance.Based on field measurement, similar material simulation and numerical simulation,combined with mechanical theory, the dissertation has made clearly of the mechanicaltheory and regularity of overlying strata and surface movement for deep mining. Amethod of surface subsidence prediction for deep mining has been built, whichprovide a technical method of building damage evaluation, the principal tasks andoutcomes are as follows:(1) An optimization method of probability integral method parameters backanalysis based on genetic algorithm (GA) was proposed, which solved the problem oflong term, and arbitrary shaped multiple working face back analysis optimization.Compared with direct search algorithms, the method applies to calculate probabilityintegral parameters of arbitrary shaped multiple working faces, and calculation withaccuracy results and high efficiency.(2) Based on research of field measurement, the dissertation has made clearly ofthe mechanical theory and regularity of overlying strata and surface movement fordeep mining: with increase of mining depth, mining subsidence has characteristics ofsubsidence slowly, long term, active period not obvious, tangent of main effect angle,critical deformation angle, and limit angle increase etc. Based on measurement data ofXuzhou Coal Mine, relationship between the subsidence character parameters,probability integral method parameters and geological and mining conditions werebuilt, which provide a basis for deep mining surface movement parameters selection.(3) In order to solve the problem of difficulty to get the reliable rock mechanicalparameters in numerical simulation experiment, intelligent mechanical parameter backanalysis method based on BP neural network were proposed, with reliable trainingsamples of BP neural network with orthogonal design method, mathematical modelwas established between mining subsidence and covered rock mechanical. The modelsolves the selection difficult of overlaying rock mechanical parameters in numericalsimulation. The research provides a basis for selection of mechanical parameters inmining subsidence numerical simulation. Example of proves that the method isrationality and reliability.(4) According to mining process, deep mining was summarized as three kinds: initial deep mining, multiple levels deep mining of inclined coal seam and multipledeep coal seams repeated mining. Covered rock and surface movement regularitywere studied based on similar material simulation and numerical simulation, influencemechanism and regularity was obtained between mining degree, coal seam dip angle,mining depth, multiple coal seam mining and rock and surface movement: miningsubsidence decrease with coal seam dip angle increase for multiple levels deep miningand surface movement limit angle increase with coal seam dip angle increase; thesubsidence increase much more than single coal seam mining, etc. These rules laid thefoundation for understanding the overlying strata and surface movement mechanismof deep mining.(5) On the basis of mechanical properties and movement characters analysis ofbedrock and alluvium, the overburden rock is divided into two parts: bed rock andloose layers, double-medium surface and rock subsidence prediction model for Deepmining was established combined with the elasticity theory. The model contains:1)elastic thin plate fixed at four sides before the first breaking of basic roof;2)simply-supported plate with winkler foundation after the first breaking of basic roof;3)3D diffusion model of non-uniform subsidence space in loose layer. The study offers anew method for the overlying rock and the surface movement calculation for deepmining.(6) Based on the basic theory of probability integral method, with direct integralto mining space, a calculation method of the maximum deformation for arbitraryshaped, changed mining thickness, multiple working faces was established, the modelsolves the calculation problem of maximum surface movement and deformation. AndMining subsidence and analysis system was developed, the software integrated theprobability integral method and the double medium rock movement prediction model,and provide a basis for deep mining subsidence prediction and analysis.Taking Xuehu coal mine as example, good results were obtained to inverse theloose layer subsidence parameters using genetic algorithm. Double mediumsubsidence space passing model was used to predict the mining subsidence, themaximum surface movement and deformation contour were got and prediction resultsagrees measured data, the results show that to calculate using double medium miningsubsidence prediction model is reliable.(7) For the characteristics of slowly moving, long cycle, mining very inadequateand prediction accuracy low of deep mining subsidence, based on the characteristics of PSO-SVM method for timing series optimization, long timing sequence miningsubsidence prediction based on PSO-SVM was established, nonlinear relationshipbetween pre-observation data and forecasting results can be got using the model, andreliable long time series mining subsidence prediction results can be got, it haspractical significance, for very inadequate deep mining, long time series surfacemovement prediction, to determine the surface dynamic subsidence process and thetime of surface stability.(8) With protective coal pillar design for Liuji township of Pangzhuang-ZhangXiaolou coal mine and7143working face mining subsidence prediction analysis ofSan Hejian coal mine in Xuzhou mining area, using to guidance for production ofdeep mining surface movement regularity and double medium model were analyzed;practice show that the conclusion and method can be used as the basis for deep miningdesign and mining subsidence prediction, good results can be got, and have practicesignificance for production guidance.