基于粗糙集与BP神经网络的底板破坏深度预测

Predication of Destroyed Floor Depth Based on Rough Set and BP Neural Networks

摘要:

基于粗糙集理论，以指标集为基础，建立了底板破坏深度影响因素的知识表达系统，提出了规则提取原则，通过粗糙集决策规则的数据处理，得出获得各因素对底板破坏深度的影响顺序为：采深、煤层倾角、工作面斜长、采厚、底板抗破坏能力、工作面是否有切穿型断层或破碎带。由于底板抗破坏能力、工作面是否有切穿型断层或破碎带这些参数难以确定，采用其余4种影响因素建立BP神经网络的预测模型，并根据建立的模型预测肥城煤田的9101和9507工作面的底板破坏深度。通过与实测结果对比，证明该网络模型的计算结果比规程提供的经验公式计算结果更接近实际。

Abstract:

Based on rough set theory and the index set, representation system of influencing factors of destroyed floor depth has been set up, and rule extraction principle has been put forward. Through data processing of rough set decision rule, factors controlling order of destroyed floor depth is mining depth, coal seam inclination angle, workface inclined length, mining thickness, coal floor antidestructive capacity; faults or broken zones. Because parameters of coal floor antidestructive capacity and faults or broken zones is difficult to determine, by using other four influencing factors, destroyed floor depths based on BP networks have been built. The destroyed floor depths of No.9101 workface and No.9507 workface in Feicheng coal field have been predicated according to the established network model. By comparing the results of neural network model and the results of empirical formula provided by national regulations with the actual measurement results, the results obtained by neural network model are closer to reality than the results of empirical formula calculation.