论文部分内容阅读
基于Fisher判别理论建立了煤巷围岩分类的Fisher判别分析(FDA)模型。选取巷道埋深、巷道跨度、采动影响系数,围岩强度、松动圈厚度和节理发育情况6个指标因子作为FDA模型的预测指标体系,以实测数据作为训练样本,获得了相应的判别函数。通过分析计算,去掉了重要性较弱的松动圈厚度1个指标,得到了4个判别函数。为了验证模型的有效性,首先利用15组实测数据作为学习样本对模型进行训练,采用回代估计法检验模型的有效性,回判的误判率为0,然后将建立的模型应用于同一地区的实际工程数据判别中,判别效果较好。结果表明,FDA模型简单、准确,是快速判别煤巷围岩分类的一种有效方法。
Fisher discriminant analysis (FDA) model of coal surrounding rock classification was established based on Fisher discriminant theory. Six index factors such as roadway depth, roadway span, mining influence coefficient, surrounding rock strength, loose ring thickness and joint development were selected as the forecasting index system of FDA model. The measured data were taken as training samples, and corresponding discriminant functions were obtained. By analyzing and calculating, one index of looseness with less importance is removed, and four discriminant functions are obtained. In order to verify the validity of the model, first of all, 15 groups of measured data were used as training samples to train the model. The effectiveness of the model was verified by the back-estimation method. The false positive rate was 0 and the model was applied to the same region The actual engineering data to determine the discriminant effect is better. The results show that the FDA model is simple and accurate, which is an effective method to quickly distinguish the surrounding rock of coal roadway.