基于随机森林法的煤矿微震危害预测

微震过程的复杂性和不均衡性导致线性模型不足以预测微震灾害,提出随机森林方法(random forest)在高能量(E≥104J)下关于煤矿开采微震灾害的预测问题.数据来自位于波兰的采用长壁开采法的煤矿,采用随机森林方法对数据集进行标准均方误差(NMSE)分析,并与决策树、Bagging算法、支持向量机、最近邻法比较,发现随机森林方法对多样本、高维度的煤矿矿山微震预测问题效果理想.

Microseismic hazard prediction in coal mines based on the random forests method

The linear model is very difficult to predict the seismic tremors due to the complexity and imbalanced of the microseismic process,The paper presents the results of the random forests method for the classification of the strong seismic mine tremors of high energy( E > 104J) of microseismic hazard sate in coal mines. The data come from the longwall mining located in a polish coal mine the using random forests method to analysis the normalization mean square error( NMSE) of the data sets,compares it with decision tree,bagging algorithm,the support vector machine and nearest neighbor method,and it concludes that the Random Forests method is ideal for the hazard prediction in coal mines with a variety of attributes and problems.