中文摘要：

导水裂隙带高度的预测对煤矿安全开采有重要意义,而传统回归方法未考虑因素间相关系数对预测结果的影响。选取采深、煤层倾角、煤层厚度、煤层硬度、岩层结构、顶板岩石单轴抗压强度、开采厚度和采空区斜长作为预测导水裂隙带高度的影响因素,建立基于PCA-BP神经网络的导水裂隙带高度预测模型。测试结果表明,煤层厚度对导水裂隙带高度的影响最大,其余各因素对导水裂隙带高度的影响较大,采深和开采厚度对导水裂隙带高度的影响较小;PCA-BP神经网络模型的训练速度和预测效果均优于BP神经网络模型,且最大预测误差仅为5.58%。

英文摘要：

The prediction of the height of water flowing fractured zone is of great importance for coal mining safety. However, traditional regression methods haven＇t considered the influence of correlation coefficients between factors on prediction performance. In this paper, the mining depth, coal seam inclination angle, coal seam thickness, coal seam hardness, rock structure, the uniaxial compressive strength of rock, mining thickness and goal plagioclase were identified as the influencing factors for height forecast of water flowing fractured zone. Then a PCA-BP neutral network was developed to forecast the height of water flowing fractured zone. Results show that the coal seam thickness has the greatest influence on the height forecast of water flowing fractured zone, while the mining depth and mining thickness have a smaller influence, and the others have a middle influence,and that the speed of convergence and prediction accuracy of the PCA-BP neutral network are both better than that of the BP neutral network with the highest prediction error of 5.58%.