基于模糊神经理论的深井煤层底板突水因素研究  

Research on main influence factors of deep seam mining floor water-bursting based on combined fuzzy neural network

作　　者：孙明[1] 张文泉[2] 郭启忠 马凯[1] 

机构地区：[1]内蒙古科技大学煤炭学院,内蒙古包头014010 [2]山东科技大学资源与环境工程学院,山东青岛266510 [3]山东新陶阳矿业有限责任公司,山东泰安271613

出　　处：《湖南科技大学学报（自然科学版）》2011年第4期5-10,共6页Journal of Hunan University of Science And Technology：Natural Science Edition

基　　金：中国博士后基金项目(20090461257);教育部春晖计划(Z2009-1-01052);内蒙古自然科学基金项目(2009MS0904)

摘　　要：深井煤层底板突水是一个复杂的水文地质力学系统,各影响因素共同作用、彼此关联和相互耦合.在"九因素学说"的基础上,笔者通过隶属函数和隶属度实现各因素数据的规范处理,选取相对保守合适的参数建立神经网络模型.选取深井回采面突水实例验证模糊神经模型,得到输入层、隐含层和输出层之间的权值系数矩阵,最终以绝对影响性系数衡量各个主控因素的贡献权重.其中,水文地质条件和煤层开采条件是造成深井煤层底板突水风险程度高的主要原因.结果表明,这种方法能够有效消除人为影响和显著增强模型动态,具有一定的研究价值和较高的实际意义.Deep seam mining floor water - bursting is a complex hydrological geomechanics system , whose influencing factors are combined action, in association and intercoupling each other. Based on ＂nine factors theory＂, every factor＇s information data were normalized by the membership function or membership grade, then to choose the comparely proper and conserve network＇ s paramater was used to build the FNN （fuzzy neural network） distinguishment model. Its reliability was tested by the engineering projects, the weight coefficient matrix of input layer, middle layer and output layer. At last, the absolute influence coefficient was to measure every main fators, the hydrological geomechanics condition and coal seam mining state was to make much risk level of deep seam mining floor water - bursting. The results showed that it effectively eliminate human impact and enhance model dynamic, so it has some research value and practical significance.

关 键 词：深井煤层底板突水 规范处理 神经网络 贡献权重 

分 类 号：TP745[自动化与计算机技术—检测技术与自动化装置]