基于LightGBM Stacking模型融合的多传感器甲烷检测系统

利用多传感器检测技术结合集成学习方法,设计了一种基于LightGBM Stacking模型融合的多传感器甲烷检测系统,其中分别使用恒压供电式催化燃烧型传感器、两路脉冲供电式催化燃烧型传感器和热传导型传感器相结合的方式达到全范围检测的目的。实验结果表明,相比使用单一算法模型,本文所提出的LightGBM Stacking集成模型在甲烷预测的准确率、均方根误差及决定系数等指标上均有更好的表现。

人工智能在长壁煤矿井下瓦斯实时预测中的应用

检测长壁开采面中爆炸性甲烷-空气混合物的形成仍然是一项具有挑战性的任务,尽管大气监测系统和计算流体动力学建模用于检查甲烷浓度,但它们不足以作为关键区域(例如靠近切削鼓)的实时预警系统,建立了长短期记忆算法,在爆炸发生前预测和管理长壁开采作业中的爆炸性气体区。介绍了一种采用人工智能算法的新方法,即改进的长短期记忆,以检测长壁开采面中爆炸性甲烷-空气混合物的形成,并在爆炸性气体成为危害之前识别可能的爆炸性气体积聚。该算法基于CFD模型输出对采煤机的六个位置进行了训练和测试,以匹配切割机的相似位置和操作条件。结果表明,该算法可以在3D中预测爆炸性气体区域,对于不同的设置,总体准确度在87.9%~92.4%;在将测量数据输入算法后,输出预测花费了2 min。结果发现,与计算流体动力学和大气监测系统相比,使用所提出的算法可以更快、更突出地覆盖准确的实时爆炸性气体聚集预测。

Methane Emissions from Grazing Holstein-Friesian Heifers at Different Ages Estimated Using the Sulfur Hexafluoride Tracer Technique

Although the effect of animal and diet factors on enteric methane （CH4） emissions from confined cattle has been extensively examined, less data is available regarding CH4 emissions from grazing young cattle. A study was undertaken to evaluate the effect of the physiological state of Holstein-Friesian heifers on their enteric CH4 emissions while grazing a perennial ryegrass sward. Two experiments were conducted： Experiment 1 ran from May 2011 for 11 weeks and Experiment 2 ran from August 2011 for 10 weeks. In each experiment, Holstein-Friesian heifers were divided into three treatment groups （12 animals/group） consisting of calves, yearling heifers, and in-calf heifers （average ages： 8.5, 14.5, and 20.5 months, respectively）. Methane emissions were estimated for each animal in the final week of each experiment using the sulfur hexafluoride tracer technique. Dry matter （DM） intake was estimated using the calculated metabolizable energy （ME） requirement divided by the ME concentration in the grazed grass. As expected, live weight increased with increasing animal age （P 〈 0.001）; however, there was no difference in live weight gain among the three groups in Experiment 1, although in Experiment 2, this variable decreased with increasing animal age （P 〈 0.001 ）. In Experiment 1, yearling heifers had the highest CH4 emissions （g·d^-1） and in-calf heifers produced more than calves （P 〈 0.001 ）. When expressed as CH4 emissions per unit of live weight, DM intake, and gross energy （GE） intake, yearling heifers had higher emission rates than calves and in-calf heifers （P 〈 0.001）. However, the effects on CH4 emissions were different in Experiment 2, in which CH, emissions （g·d^-1） increased linearly with increasing animal age （P 〈 0.001）, although the difference between yearling and in-calf heifers was not significant. The CH4/live weight ratio was lower in in-calf heifers than in the other two groups （P 〈 0.001 ）, while CH4 energy output as a proportion of GE intake was lower in calves than in yearling and in-calf heifers （P 〈 0.05）. All data were then pooled and used to develop prediction equations for CH4 emissions. All relationships are significant （P 〈 0.001）, with R2 values ranging from 0.630 to 0.682. These models indicate that CH4 emissions could be increased by 0.252 g.d-1 with an increase of I kg live weight or by 14.9 g·d^-1 with an increase of 1 kg·d^-1 of DM intake; or, the CH4 energy output could be increased by 0.046 MJ·d^-1 with an increase of 1 MJ·d^-1 of GE intake. These results provide an alternative approach for estimating CH4 emissions from grazine dairy heifers when actual CH, emission data are not available.

Prediction on adsorption ratio of carbon dioxide to methane on coals with multiple linear regression

The multiple linear regression equations for adsorption ratio of CO2/CH4 and its coal quality indexes were built with SPSS software on basis of existing coal quality data and its adsorption amount of CO2 and OH4. The regression equations built were tested with data collected from some s, and the influences of coal quality indexes on adsorption ratio of CO2/CH4 were studied with investigation of regression equations. The study results show that the regression equation for adsorption ratio of CO2/CH4 and volatile matter, ash and moisture in coal can be obtained with multiple linear regression analysis, that the influence of same coal quality index with the degree of metamorphosis or influence of coal quality indexes for same coal rank on adsorption ratio is not consistent.

Prediction of methane emissions during the extraction of close-to-roof layer of a seam of large thickness

Prediction of methane emissions at the stage of longwall planning constitutes the basis for the determination of the appropriate method and parameters of ventilation and selection of prevention means including the methane drainage technol- ogy. The growth of methane saturation of coal seams with the extraction depth, with simultaneously increasing output concen- tration, contributes to the increase of the quantity of methane emitted into longwall areas. The subject matter of the article has been directed at the predicted quantity of methane emissions into planned longwalls with roof caving in the layer of seams adjacent to the roof of large thickness. The performed prognostic calculations of methane emissions into the longwall working were referred to two sources, i.e. methane liberated during coal mining by means of a cutter-loader and methane originating from the degasification of the floor layer destressed by the longwall conducted in the close-to-roof layer. The calculations of predictions allow to refer to the planned longwall, on account of the emitting methane, with possible and safe output quantity. Planning of extraction in the close-to-roof layer of a seam of large thickness with roof caving is especially important in con- ditions of increasing methane saturation with the depth of deposition and should be preceded by a prognostic analysis for de- termining the extraction possibilities of the planned longwall.

Stochastic prediction and control to methane in coalmine

The ventilation system plays an essential role in underground workings, and improvements in dilution effect to stochastic methane build-up at cul-de-sac of a coalmine require the installation of mixed ventilation system. For 4-12-1 I N02.8A centrifugal ventilation fan, the characteristic operating function of its mixed ventilation system is calculated from ventilation quantity and total pressure in the actual working status. At cul-de-sac of the reference coalmine, the evolution of methane concentration is a compound Poisson process and equivalent to a Brownian motion for Gaussian distributed increments. Solution of stochastic differential equation driven by mixed ventilation system, with dilution equation for its closure, provides parameters of mine ventilation system for keeping methane concentration within the permissible limit at cul-de-sac of the reference coalmine. These results intend to shed some light on application of blowing-sucking mixed ventilation systems in underground workings, and establish stochastic trends to consider methane control in coalmines.

Theoretical predictions of viscosity of methane under confined conditions

Density functional theory has been confirmed as a reliable approach in the descriptions of inhomogeneous fluids.By integrating the density functional theory into the revised local average density model, a theoretical approach is constructed to investigate the local shear viscosity in the confined conditions. In the density functional theory,the weighted density approximation for attractive part and the modified fundamental measure theory for repulsion contribution are adopted to accurately describe the inhomogeneous systems. By comparing with simulation data, the theoretical model is tested. In this work, the shear viscosities of methane are calculated in different external fields(on a hard wall, a solvophobic wall and in slit pores with different widths). In addition, the effects of temperature on the local density and viscosity are also considered. It shows that the effect of temperature on the shear viscosity is more obvious on solid surfaces. The calculation provides an approach to determine the viscosity under confined conditions, which is extremely significant in real industrial applications.