基于BP神经网络的导水裂隙带高度预测

Forecast of the height of water flowing fractured zone based on BP neural networks

为准确预测工作面导水裂隙带发育高度,在总结顶板导水裂隙带高度预测方法和理论的基础上,结合大量实际资料,分析归纳出采深、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是影响导水裂隙带高度的主要因素。根据全国典型案例,建立了基于BP神经网络的导水裂隙带高度预测模型,确定了BP神经网络所需的输入样本和测试样本,运用Matlab软件对网络进行了训练,得到了优化的网络模型,并利用建立的模型预测了焦作煤田赵固一矿11011工作面导水裂隙带高度。通过与实测结果对比,证明基于BP神经网络建立的导水裂隙带高度预测模型的计算结果比规程提供的公式计算的结果更接近实际。

In order to predict exactly the forming height of water flowing fractured zone, based on summing up methods and theories to forecast the height of water flowing fractured zone, combined with a lot of practical information analysis, the main eight factors influencing the height of water flowing fractured zone have been found out, which are mining depth, coal seam inclination angle, coal seam thickness, coal seam hardness, rock structure, the uniaxial compressive strength of rock, mining thickness and goaf plagioclase. According to the typical cases in China, the forecasting model of the height of water flowing fractured zone based on BP neural networks has been built, and the input samples and testing samples needed by BP neural networks have been determined. The optimization of network model has been got after Matlab software is used in the training of network. The height of water flowing fractured zone of 11011 workface in Zhaogu No. 1 Mine, Jiaozuo cola field has been forecasted according to the established network model. Compared with measured results, it has shown that the calculation results obtained by the BP neural network model are closer to the reality than those obtained according to the empirical formula.