Study On Main Controlling Factors And Prediction Models Base On GA-WNN Of Mining Subsidence Coefficient

According to gological conditions of Shendong mining area, the main controlling factors of mining subsidence are selected and introduced into the models of mining subsidence prediction by means of theoretical analysis, numerical simulations and scientific programs. And the relation between subsidence coefficient and the main controlling factors are derived; the Optimized Wavelet Neural Network based on Genetic Algorithm (GA-WNN) is applied to the prediction of mining subsidence.Using improved fuzzy analytical hierarchy process, the master-factors of mining subsidence are selected as the ratio and thickness of losses bed, mining thick, comprehensive hardness of cover rocks, the position and type of key stratum, each of whose weight are 0.2211,0.1538,0.1489,0.1138,0.0861.Under the full minnig of the same intensity, subsidence coefficient is inversely proportional to comprehensive hardness of cover rocks, and proportional to the ratio and thickness of losses bed. When the comprehensive hardness of cover rocks is difficult to determine, the subsidence coefficient (η) is predicted using the ratio (λ) and thickness (χ) of losses bed, the regression equation is as followes:Based on GA-WNN and numerical simulations, the other model of mining subsidence prediction is conducted. The practical simulation results show that GA-WNN model can effectively increase the diagnostic accuracy, which results are close to actual experiences. The model applies to prediction of mining subsidence.