基于BP人工神经网络的封闭不良钻孔导水危险性评价

摘要

In this PaPer,a comPrehensive analysis was made on the major factors affecting the water inrush risk of the badly-sealed boreholes and the evaluation indicators of each factor,and an evaluation indicator system for the water inrush risk of the badly-sealed boreholes was set uP by taking some badly-sealed boreholes in Liangjia Mine as the study object. There are three first-grade indicators:they are the water filling source,geological structure and the influence of mining activities,among them,the water filling source contained three secondary-grade indicators,and the geological structure and the influence of mining activities resPectively contained two indicators. Based on the established indicator system for evaluating the badly-sealed boreholes and by using MATLAB software,a three-layer BP artificial neural network model was built for quantitatively evaluating the water inrush risk of the badly-sealed boreholes. The inPut layer of the model has three nodes:i. e. the influence index of the water filling source,the influence index of the geological structure and the influence index of the mining activities. The middle layer has nine nodes and the outPut layer has one node,i. e. the water-inrush risk index of the badly-sealed boreholes. Based on the evaluation results,the water inrush risk of the badly-sealed boreholes was divided into three grades:safer,more dangerous and very dangerous.%以梁家煤矿部分封闭不良钻孔作为研究对象，综合分析了影响封闭不良钻孔导水危险性的主要因素以及各因素的评价指标，建立了封闭不良钻孔导水危险性评价指标体系。一级指标有3个，分别为充水水源、地质构造、采掘活动的影响。其中，充水水源下设有3个二级指标，地质构造和采掘活动的影响分别下设有2个二级指标。根据已建立的封闭不良钻孔评价指标体系，运用MATLAB软件构建了3层BP人工神经网络模型，对封闭不良钻孔的导水危险性进行定量评价，模型输入层有3个节点，分别为充水水源影响指数、地质构造影响指数、采掘活动影响指数。中间层有9个节点，输出层有1个节点即封闭不良钻孔导水危险性指数，根据其评价结果划分出钻孔的导水危险性等级，分别为较安全、较危险和危险。