基于PCA-FDA模型的顶煤冒放性评估研究

Evaluation of top coal cavability based on PCA-FDA model

为合理评估顶煤冒放性、判定煤层能否进行综放开采,基于主成分分析(PCA)与Fisher判别分析(FDA)构建了PCA-FDA的顶煤冒放性评估模型。该模型选取煤层赋存深度、煤体单轴抗压强度、煤层夹矸层厚度、煤层厚度、煤层节理裂隙发育程度和直接顶充填系数6项指标作为顶煤冒放性评估的样本变量,运用PCA消除变量间重叠信息,提取了4个主成分作为FDA的判别因子,并运用FDA对我国23个放顶煤开采工作面实测数据进行学习和训练。同时,利用回代估计法对模型进行检验,正确率为91.30%,具有较高的精确度。将建立的模型应用于山西省的2个工程实例,结果表明:该模型科学合理、结果可靠,为顶煤冒放性评估和放顶煤工艺参数选择提供了一种科学的参考方法。

In order to evaluate top coal cavability reasonably,and determine whether coal seam is suitable for top-coal caving. The top coal cavability evaluation model( PCA-FDA) was established by combining principal component analysis( PCA) and Fisher discrimination analysis( FDA). Six influential factors were selected as sample variances for top coal cavability evaluation including the coal seam depth,uniaxial compressive strength of coal,coal thickness of seam folder gangue layer,thickness of coal seam,the development degree of the coal seam joint and fractures and coefficient of direct top filling. PCA was used to eliminate overlapping information in the variables,thus four principal components are extracted as the discriminant factors of FDA. FDA is applied to learn and train measured data of 23 top-coal caving mining face in China. Meanwhile,back substitution estimation method was adopted to verify the model effect,and accuracy is very high up to 91.30%. Finally the established evaluation model was applied in two engineering projects in Shan'xi province,and results show that this evaluation method is scientific,reasonable and reliable. The model can provide a scientific reference for evaluation of top coal cavability and technology parameters selection in top coal caving.