矿井瓦斯超限实时预警技术研究

Research on gas overrun real-time early warning technology in mine

为解决瓦斯超限的预测预报问题,根据瓦斯超限前煤体及工作面环境表现出的前兆特征,依据矿井瓦斯超限影响因素,从瓦斯、通风、地质构造、管理等方面,建立了瓦斯超限预警指标体系,利用瓦斯体积分数变化指标作为实时预警指标,其他指标作为趋势预警指标,针对瓦斯超限单指标误报率高的问题,利用模糊数学理论建立了瓦斯超限多指标模型,确定了权重值和评价值。在某矿井建立了工作面瓦斯超限实时预警系统,输出红色、黄色及绿色3级瓦斯超限预警结果,得出了瓦斯超限发生的概率,根据瓦斯超限预警结果采取相应措施可以大幅降低瓦斯超限次数;多指标模型预警与单指标模型预警对比结果表明:红色预警次数可以减少50%以上,黄色预警次数减少30%以上,说明采用多指标量化模型比单一指标模型更为合理,提高了预警准确率,为煤矿安全生产提供管理依据,也为类似预警指标研究提供参考。

In order to solve the problem of forecasting gas overrun,according to the precursory characteristics of coal body and working face environment before gas exceeds the limit,the index system of gas overrun early warning was established from the aspects of gas,ventilation,geological structure and management and other influencing factors,using the gas,the influencing factors of gas overrun in a coal mine,volume fraction change index as a real-time early warning indicator,and other indicators are used as trend early warning indicators.Based on the problem of high false alarm rate of single indicator of gas exceeding limit,a multi-index model of gas exceeding limit is established by using the theory of fuzzy mathematics,and the weights and evaluation values are determined.A real-time warning system for gas overrun in working face are established in a affiliated coal mine,and the three-level warning results of red,yellow and green gas overrun were output,and the probability of occurrence of gas overrun are obtained.According to the results of gas overrun warning,the corresponding measures can be taken and this can greatly reduce the number of gas overrun.The comparison between multi-index model warning and single-index model warning is made:the number of red warning times can be reduced by more than 50%and the number of yellow warning by more than 30%,indicating the multi-index quantitative model is more reasonable than the single-index model,which improves the accuracy of early warning,provides management basis for coal mine safety production and supply references for similar early warning index research.