曲线拟合与分段函数组合法在瓦斯抽采量预测中的研究与应用

Research and Application of Curve Fitting and Segmentation Function Combination Method in Gas Extraction Prediction

为减少矿井煤与瓦斯突出、瓦斯爆炸等事故的发生,将井下瓦斯及时抽出是必要的,因此对瓦斯抽采量进行准确地预测愈发重要。在神经网络法、多元线性回归法等预测方法的基础上,根据瓦斯抽采量在时间效应下变化波动大的特性,提出了曲线拟合与分段函数组合法。此方法以瓦斯抽采量衰减为界将瓦斯抽采分为前、后2个阶段,根据实测数据作抽采纯量随时间变化曲线并对变化曲线进行拟合,根据拟合曲线列出预测瓦斯抽采量的计算公式,用于对抽采量进行预测。分别对九里山矿、古汉山矿进行后段抽采量预测与实际的比较,其中九里山矿得到后段预测值的最小相对误差为0.63%,最大相对误差为2.34%,古汉山矿得到后段预测值的最小相对误差为0.09%,最大相对误差为3.23%。表明曲线拟合与分段函数组合法对井下瓦斯抽采量预测是可行的,对井下瓦斯治理具有重要的意义。

In order to reduce the occurrence of accidents such as coal and gas protrusion and gas explosion in mines,it is necessary to extract the underground gas in time,so it is more and more important to make accurate prediction of gas extraction amount.Based on the prediction methods such as neural network method and multiple linear regression method,the combination of curve fitting and segmentation function method is proposed according to the characteristics of the gas extraction amount which fluctuates greatly under the time effect.This method takes the decay of gas extraction as the boundary to divide the gas extraction into two stages,before and after,and makes the extraction pure volume change curve with time according to the measured data and fits the change curve,and lists the calculation formula for predicting the gas extraction volume according to the fitted curve,which is used to predict the extraction volume.Comparison of predicted and actual back-end pumping volumes for the Jiulishan and Guhanshan mines.The minimum relative error of the predicted amount of back-section extraction was 0.63%and the maximum relative error was 2.34%for Jiulishan mine,respectively,and the minimum relative error of the predicted amount of back-section extraction was 0.09%and the maximum relative error was 3.23%for Guhanshan mine.It shows that the combination of curve fitting and segmentation function method is feasible for the prediction of underground gas extraction,which is of great significance for underground gas management.