煤层注水难易程度的BP神经网络评价法  

BP neural network method application in feasibility evaluation of water infusion for coal seam

作　　者：秦书玉[1] 秦威[2] 赵景馥[1] 

机构地区：[1]辽宁工程技术大学电子与信息工程系,辽宁阜新123000 [2]北京航空航天大学机械工程及自动化学院,北京100083

出　　处：《中国地质灾害与防治学报》2006年第3期87-90,共4页The Chinese Journal of Geological Hazard and Control

摘　　要：科学评价煤层注水难易程度,既是判定煤层是否可注及可注效果,也是确定煤层注水方式和工艺参数的重要依据。但是,由于煤层注水难易性受多种因素影响,属于极为复杂的非线性问题,很难用传统的数学方法进行定量评价。文献[1]用模糊聚类方法进行定量评价,在实际应用中,由于精度不够,出现过误判。为此,针对影响评价煤层注水难易程度的多种因素,如煤层孔隙率、煤的坚固系数、湿润边角、饱和水份增值等所构成的复杂系统,采用具有非线性映射功能的神经网络评价煤层注水难易程度,将煤层注水难易程度分为4级。并将该方法的评价结果同模糊聚类进行了对比,通过比较和实例验证,其误差为5.6%,提高了10个百分点。此方法评价精度较高,可用于指导实际煤层注水工程。Evaluation of water infusion for coal seam is very important not only for judging whether the coal seam can be injected and its consequence, but also for conforming coal seam water flooding regime and technica parameters. But coal seam water infusion is a non-linear problem with extremely complexity, and it can hardly be solved by traditional mathematical methods because it is influenced by many factors. Fuzzy clustering method in reference [ 1 is used for quantitative evaluation], would lead to wrong result in practice because of its low accuracy. Therefore, neural network with non-linear mapping function is adapted to evaluate feasibility of coal seam water infusion, using seam porosity, firm coefficient of coal, wet rim angle, saturated moisture increment and so on. The feasibility of coal seam water infulsion thus was marked into four grades. Compared with the results of evaluation in fuzzy clustering method, the error of neural network evaluation method is 5.6% and 10 percentage points higher than fuzzy clustering method. The method applied in this paper has higher evaluation accuracy and can be used for guiding the actual coal seam water infusion engineering.

关 键 词：煤层注水 神经网络 评价方法 影响因素 模糊聚类方法 

分 类 号：TP39[自动化与计算机技术—计算机应用技术] TD713[自动化与计算机技术—计算机科学与技术]