论文部分内容阅读
针对煤层底板突水问题,提出了基于主成分分析、模糊数学和随机森林的一种新预测模型。首先通过主成分分析将6个影响因素(水压、采高、隔水层厚度、断层落差、煤层倾角、断层距工作面距离)进行降维,提取4个主成分因子,其次对主成分因子进行模糊化,作为随机森林模型的输入变量,建立基于PCA_Fuzzy_RF的煤层底板突水预测模型。利用华北矿区实测资料的50组数据作为PCA_Fuzzy_RF模型的训练数据,10组数据作为测试数据,并将预测结果与BP神经网络及Fisher模型进行对比分析,结果表明,PCA_Fuzzy_RF模型的误判率为0,适用于解决煤层底板突水问题。
Aiming at the problem of water inrush from seam floor, a new prediction model based on principal component analysis, fuzzy mathematics and stochastic forest is proposed. Firstly, principal component analysis was used to reduce the dimensions of six influencing factors (water pressure, mining height, thickness of aquifuge, fault drop, coal seam dip and fault distance to working face), and four principal component factors were extracted. Secondly, As the input variables of stochastic forest model, a prediction model of coal floor water inrush based on PCA_Fuzzy_RF is established. 50 sets of measured data in North China Mining Area were used as training data of PCA_Fuzzy_RF model and 10 sets of data were used as test data. The results of the prediction were compared with BP neural network and Fisher model. The results showed that the false positive rate of PCA_Fuzzy_RF model was 0, Suitable for solving coal floor water inrush problem.