Studies On Prediction Technology Of The Mining Surface Movement And Deformation

In mine subsidence monitoring,surface subsidence prediction is an important dynamic content. To further guide rail,the next building,mining work under water,we need some time to forecast the future by moving the deformation of the surface caused by underground coal mining.This paper details the observation situation Panyi mine east pit1252(1) face natural and geographical conditions,geological and mining conditions,stations and stations after layout by establishing the BP neural network prediction model,Kalman filtering prediction model and Correct the error filtering prediction model of Calman for Panyi mine east pit1252(1) face toward the line of continuous dynamic point elevation forecasts.At present many neural network models,BP neural network is the most widely used at present,is a kind of multilayer feedforward and training network according to the error back propagation algorithm.Detailed research on the basis of BP neural network is presented in this paper, using the theory of the working face to line continuous point elevation for dynamic forecast, and the comparative study on the prediction and measured data analysis,data show that when the sinking speed larger monitoring forecast, forecast error is bigger; In general speaking, the BP neural network prediction theory to meet the needs of the mining surface movement dynamic forecast.Kalman filter method is a kind of linear recursive filtering method, is the use of state now predicted value and the present state of measurement values appear in state estimation, kalman filter prediction model is the use of the present stage of the kalman filter optimization estimates to forecast the future state.By using this theory to forecast to the continuous point elevation, it is concluded that when monitoring with subsidence mutation occurs, working face advance prediction error is bigger.In order to reduce the prediction error, according to theforecast error curve gives Correct the error filtering prediction model of Calman, data shows that the model error correction proposed indeed relatively reduce the forecast error.In order to improve the operation efficiency, based on Visual Basic6.0as development platform, to develop the mining surface movement prediction system, to realize the automatic forecast.