【作者】 李雨成； 刘天奇； 周西华；

【Author】 LI Yu-cheng;LIU Tian-qi;ZHOU Xi-hua;College of Safety Science and Engineering,Liaoning Technical University;Key Laboratory of Mine Thermo-motive Disaster and Prevention,Ministry of Education;

【机构】 辽宁工程技术大学安全科学与工程学院； 矿山热动力灾害与防治教育部重点实验室；

【摘要】 为提高矿井煤尘爆炸预警性能,促进煤矿安全生产便于及时采取煤尘防爆措施,本文选用40种煤样,利用实验室相关仪器按照相关国家规定,对影响煤尘爆炸性的35项因素进行测定。为考虑分散度的影响作用,首次定义煤尘粒径分散系数。由于煤尘爆炸影响因素相关性诊断检验统计量KMO值高达0.732,所以使用主因子分析法对变量进行降维,同时增加了爆炸预警结果准确度。得到影响煤尘爆炸的五大主因子得分,结合爆炸火焰长度值进行BP网络训练,通过使用附加动量法提高网络学习效率。虽然10种检验煤样预测结果中最大误差达到10.1%,但总体预测有效率达到90%,说明该预警方法具有一定科学性与适用性。该结论对促进煤矿安全生产具有一定指导作用,同时对煤尘爆炸安全预警工作具有一定借鉴意义。更多还原

【Abstract】 To improve the prediction performance of coal dust explosibility in mine,and to promote the safe production of coal mine that it can take explosion-proof measures in time,in this paper,40 kinds of coal samples are selected to condust relevant experiments in accordance with relevant state regulations with laboratory equipments.35 kinds of factors are measured which can affect the explosibility of coal dust.To reflect the coal dust dispersion characteristics,dust particle size dispersion coefficient is defined for the first time.Since dust explosive factors associated diagnostic test statistics KMO value is up to 0.732,the method of principal factor analysis is used to reduce the dimension of the variable,which can increase the explosion predicting accuracy at the same time.The five main factors’ score of 40 coal samples which can affect the explosion is got.Combined with flame length of coal dust explosion,the BP network is trained.The additional momentum method is used to improve the network.Although the maximum predicting error of 10 tested coal samples is up to 10.1%,the overall predicting efficiency reaches 90%,which indicating that the predicting method has certain scientificalness and applicability.The conclusion has some guidance on the promotion of safety in coal mine production,meanwhile it has certain reference meaning to safety warning in coal dust explosion.更多还原