Mine Pressure Prediction Study Based On Wavelet Analysis And ARMA Combination Model

Mine pressure acts on the mining face periodically during the coal production, which impacts on coal mine production safety and even leads to mine accidents. Thus, the study on the pressure behavior and its intensity of the mining face has important theoretical and practical significance to the safety production in the coal mining process.This paper reviewed the relevant research both home and abroad, conducting a systematic study on the underground pressure forecasts through literature analysis, theoretical analysis and field research and comprehensive usage of mine pressure behavior theory, statistical analysis, wavelet analysis, time series analysis, qualitative and quantitative analysis. The main work and conclusions are as follows:(1) This paper analyzes the study status on the mine pressure behavior prediction, the conclusions are: The mechanical method based on the mine pressure theory is the most widely used; The probabilistic methods are also used in pressure criterion and dynamic load factor; The newly emerging smart technology methods aiming at nonlinear issues in mine pressure behavior forecasting are seldom used(2) The paper takes 20307 Coal Seam of Nan Liang Coal Enterprise as an example, according to the statistical results, the paper gets the following conclusions: The periodic weighting criterion for 20307 Coal Seam is 24.4MPa; The pressure intensity of the immediate roof is 24.86MPa; The pressure step of the immediate roof is 10.36m; The initial pressure intensity of the main roof is 31.68MPa; The periodic pressure step of the main roof is 22.5m and the pressure intensity is 37.1MPa.(3) The paper builds the Wavelet Denoising and ARMA Combined Model, as well as Mallat Algorithm and ARMA Combined Model according to the analysis of the roof weighting factors. Then the paper further discusses the feasibility of the two proposed combined models in the mine pressure behavior prediction.(4) The paper applies the two proposed combination models in a mine in northern Shaanxi and gets the following results according to the model error analysis: The combination model is better than the single model. Both of the two Combined Models have encountered relatively accurate predictions of bracket resistance, periodic weighting step distance and the promote distance. And, Mallat algorithm and ARMA Combined Model has a relatively higher accurecy than the other Combined Model according to the error analysis, which can better meet the security needs of the production. Meanwhile, extracting each factor has a further influence on the accuracy of the roof pressure and the promote distance prediction than de-noising. Thus, extracting various factors should be firstly considered in such studies, followed by is the noise signal.The results of this study can offer some reference value for grasping the Mine pressure rule and Periodical Pressure Forecast Research, providing a reference method for mine pressure behavior prediction.