基于正交试验-多元回归的煤矿瓦斯涌出量预测

Prediction and forecast of the coal mining gas emission based on the orthogonal test and multiple regression

为了快速准确地预测工工作面瓦斯涌出量,简化瓦斯涌出量预测建模过程,将正交设计方法与多元回归分析相结合提出了简化的预测模型,并针对实际矿井数据进行应用。结果表明,结合正交设计中的极差和方差分析可以有效剔除对瓦斯涌出量影响不显著的因素,该剔除结果与SPSS软件显著性分析结果一致,利用SPSS软件对剩余的瓦斯涌出量影响因素进行多元回归分析得到相应的回归模型,经检验其预测结果相对误差平均值为1.76%,回归模型具有较高的精确度,可为煤矿瓦斯涌出量预测工作提供一定的参考。

The given paper intends to propose a simplified forecasting model by combining the orthogonal design method with the multivariate regression analysis and the orthogonal design steps.In order to predict the gas emission in the working face fleetly and accurately.And,to achieve the said purpose,we have first of all been trying to manage to determine the actual level value of each influential factor of the gas emission via the subsection statistics.And,then,we have managed to work out a proper orthogonal table and fill it in the gas emission data in accordance with the individual testing schedule.And,next,3 groups of the data we have worked out can be suggested and recommended at least to calculate and work out the average value before filling in the above mentioned table to get rid of the random errors.And.last of all,all the influential factors can be illustrated with a small amount of the experimental data on the screen through the range analysis and ANOVA.And,consequently,the range analysis of the orthogonal test may help us to examine all the influential factors on the change of the gas emission,the coal seam burial depth,the coal-mining intensity,the original gas content of the coal seam,the mining yield rate,the inclining angle of the coal seam,and,even the coal thickness of the adjacent layers.At the same time,we have also determined that the optimized synthesis of the factors should be the minimalist gas emission to the depth of the buried coal seam,which ought to be 400 m underneath,with the mining intensity being1500 t/d and the original gas content of coal seam being 2.5 m^(3)/t.Thus,the consequential analysis for the mining yield rate has to be at about 94%,with the thickness of the adjacent layers being 1.45 m,as well as with the inclination angle of the coal seam being 5°.And,so,fruitfully,the F ratio of ANOVA results at a different confidence rate,the confidence of the primary and secondary layers may help to verify the rest 2 factors further,which can be expected to get eliminated for they wouldn’t have any significant impact on the coal production efficiency.And,later,the elimination results we have gained can help to prove to be consistent with the analysis results of the SPSS significance.And,so,the multivariate regression analysis of SPSS gas can actually help us to find a model of the residual influential factors of the gas emission,in addition to the average relative error of its prediction results through testing,which proves to be equal to1.76%.Hence,as compared with the currently adopted forecast ones,the method we have proposed turns out to be more convenient in demanding less data and simpler in the process modeling.What is more,it can also help us to heighten the forecasting accuracy through the easier and simpler calculation.