Research And Development Of The Effective Radius Prediction System For Mine Gas Drainage Based On Deep Neural Network

The determination of the effective radius of drainage directly affects the effect of coal seam gas drainage in mines.Before the coal seam is mined,coal seam gas pre-drainage is required.Coal seam gas pre-drainage requires drainage holes to be arranged in the coal seam,and reasonable drilling spacing should be considered in the arrangement of drainage holes.Unreasonable spacing of drainage holes will not only affect the effect of gas pre-drainage,but also may cause gas drainage blank zones and gas disaster accidents,and at the same time increase unnecessary waste of manpower,material resources,financial resources,and time.At present,although researchers have proposed a variety of methods for measuring the effective radius of extraction,the measurement operation process is too cumbersome,the universality of measurement is poor,and these methods are not conducive to the use of actual coal miners on site.Therefore,in order to improve the current situation,this paper adopts a combination of basic theoretical analysis,numerical simulation research,neural network model construction,and software research and development to carry out research on determining the effective radius of reasonable extraction quickly and accurately.According to the established mathematical model,a numerical simulation is carried out to study the variation law of the effective radius of extraction affected by different factors.Using the superior prediction function of the deep neural network,the prediction index of the effective radius of the extraction was selected,and a prediction model of the effective semi-diameter of the extraction was built.Using related technologies based on Py Qt5 and other development software,a set of effective radius prediction system for extraction is developed.The research results obtained in this paper have certain application value for quickly determining the gas drainage radius on site.The main results obtained in this paper are as follows:(1)Based on the theory of coal seam gas flow and coal body deformation,the gas-solid coupling mathematical model of the drainage borehole was established,and the three-dimensional single-hole drainage was simulated by numerical simulation software.The influence of stress,drainage time,borehole diameter,and negative drainage pressure on the effective radius of drainage,it is obtained that the initial permeability of the coal seam has the greatest influence on the effective radius of drainage,and the initial coal seam gas pressure,in-situ stress,drainage time,and borehole The influence of diameter is second,and the influence of negative pressure of extraction is small,and the influence of negative pressure of extraction can be ignored.(2)By comparing and analyzing the corresponding applicable advantages of the common deep neural network(DNN),convolutional neural network(CNN),and recurrent neural network(RNN)deep learning algorithms,it is determined that the use of deep neural network deep learning algorithms is used to predict extraction.Effective radius.Through the analysis of the numerical simulation results and the conclusions obtained by consulting a large number of documents on How Net,combined with the relevant theoretical knowledge learned by the deep neural network,the input eigenvalues and output target values of the effective radius of prediction and extraction are determined.Then,the advantages and applicable scenarios of stochastic gradient descent(SGD),batch gradient descent optimization(BGD),Ada Grad algorithm,RMSProp algorithm and Adam algorithm were compared and analyzed,and the use of Adam optimization algorithm was determined,and an optimization depth based on Adam algorithm was proposed.A neural network(Adam+DNN)method for predicting the effective radius of extraction.By using the Adam+DNN optimal prediction model saved after training to predict the effective radius of extraction on the sample test set and the data collected on site,the prediction results show that the effective radius of extraction predicted by the prediction model has a high accuracy rate,and the prediction The steps are easy to understand and operate,and the prediction efficiency is high.(3)A set of effective radius prediction system for mine gas drainage based on deep neural network was developed by using related technologies based on Py Qt5 and other development software and combined with the prediction model built above.There are 35 figures,7 tables,and 115 references in this paper.