基于PCA-RF模型的冲击地压预测  

Prediction of rock burst based on PCA-RF model

作　　者：郭浩 杨超宇[1] GUO Hao;YANG Chaoyu(School of Economics and Management,Anhui University of Science&Technology,Huainan 232000,China)

机构地区：[1]安徽理工大学经济与管理学院,安徽淮南232000

出　　处：《哈尔滨商业大学学报（自然科学版）》2023年第4期413-418,501,共7页Journal of Harbin University of Commerce：Natural Sciences Edition

基　　金：国家自然科学基金(No.61873004,多源传感器环境下基于异构特征信息融合的行为识别)。

摘　　要：为了提高冲击地压危害预测的准确性,基于主成分分析法和随机森林算法,构建了由主成分分析方法优化的随机森林模型,分析静态冲击地压数据并处理异常值,通过数据标准化处理、计算相关系数矩阵及累计方差贡献率,提取出5个主要特征.利用优化的随机森林模型训练冲击地压数据集,使模型参数不断得到优化.以混淆矩阵中的准确率作为评估指标,将优化的随机森林模型与单一随机森林模型进行比较分析.实验结果表明,优化的随机森林模型比单一随机森林模型具备更好的效果,其准确率达到了88.9%,高于单一的随机森林模型,进而得出结论,即优化的随机森林模型能有效地对冲击地压危害进行预测,通过随机森林优化模型,一定程度上有效缩减冲击地压危害预测的时间.In order to improve the accuracy of rockburst hazard prediction,based on the principal component analysis method and random forest algorithm,a random forest model optimized by the principal component analysis method was constructed.The static rockburst data was analyzed and the abnormal values were processed.Five main features were extracted through data standardization,calculation of correlation coefficient matrix and cumulative variance contribution rate.The optimized random forest model was used to train the rockburst data set,so that the model parameters are continuously optimized.Taking the accuracy rate in the confusion matrix as the evaluation index,the optimized random forest model is compared with the single random forest model.The experimental results show that the optimized random forest model has a better effect than the single random forest model,and its accuracy rate reached 88.9%,which was higher than the single random forest model.The results showed that the optimized random forest model could effectively predict the rockburst hazards.Through the random forest optimization model,the time for predicting rockburst hazards was effectively reduced to a certain extent.

关 键 词：主成分分析 随机森林 冲击地压 混淆矩阵 机器学习 分类预测 

分 类 号：TD324[矿业工程—矿井建设]