基于Adam优化深度神经网络快速确定瓦斯抽采半径*

为解决煤矿瓦斯有效抽采半径难以快速准确确定的问题,采用基于Adam算法优化DNN（深度神经网络）方法来预测瓦斯抽采半径。查阅文献共收集已得到验证的970组数据集，每组数据选取煤层瓦斯初始渗透率、钻孔直径、抽采时间、地应力、煤层初始瓦斯压力作为预测模型的5个特征量，有效抽采半径作为目标输出值。接着预测模型进行不断学习和训练，最终训练得到1个最优的瓦斯有效抽采半径预测模型。利用训练好的最优预测模型结合Python语言开发出计算有效抽采半径的软件，并使用该软件在四季春煤矿和鹤煤六矿进行有效抽采半径预测的工程实例研究，验证该软件预测抽采半径的实用性和准确性。研究结果表明:通过使用开发的软件，可快速且较准确地计算出矿井瓦斯有效抽采半径，可为暂不具备现场测试条件的矿井抽采设计提供一定的参考依据。

Quickly determining gas drainage radius based on Adam optimized deep neural network

In order to solve the problem that it is difficult to quickly and accurately determine the effective gas drainage radius of coal mines,an optimized DNN (deep neural network) method based on Adam algorithm was used to predict the gas drainage radius.A total of 970 groups of data sets that had been verified were collected by referring to the literature.For each group of data,the initial permeability of coal seam gas,borehole diameter,drainage time,in-situ stress,and initial coal seam gas pressure were selected as 5 characteristic quantities of the prediction model,and the drainage effective radius was taken as the target output value.The prediction model was continuously learned and trained,and finally an optimal prediction model of gas drainage effective radius was obtained by training.Using the trained optimal prediction model combined with Python language,a software for calculating the drainage effective radius was developed,and the software was used to conduct the engineering case study on the prediction of drainage effective radius in Sijichun coal mine and Hemei No.6 coal mine,and the practicability and accuracy of the software for predicting the drainage radius were verified.By using the developed software,the drainage effective radius of mine gas could be calculated quickly and accurately,which provides a certain reference for mine drainage design without the field test conditions temporarily.