PCA-SVM模型在煤层瓦斯涌出量预测中的应用  

Application of PCA-SVM Model in Prediction of Coal Seam Gas Emission

作　　者：李鑫灵 袁梅[1,2,3] 敖选俊 隆能增 张平 许石青 LI Xinling;YUAN Mei;AO Xuanjun;LONG Nengzeng;ZHANG Ping;XU Shiqing(Mining College of Guizhou University,Guiyang 550025)

机构地区：[1]贵州大学矿业学院,贵阳550025 [2]贵州省非金属矿产资源综合利用重点实验室,贵阳550025 [3]复杂地质矿山开采安全技术工程中心,贵阳550025 [4]贵州中纸投资有限公司,贵州盘州553537

出　　处：《工业安全与环保》2019年第10期35-39,共5页Industrial Safety and Environmental Protection

基　　金：贵州省科技计划项目(黔科合支撑[2018]2789)

摘　　要：针对煤层瓦斯涌出量影响因素众多且各因素间呈复杂非线性的特点,文章利用主成分分析法(PCA)和支持向量机(SVM)的理论基础,构建了PCA-SVM的煤层瓦斯涌出量预测模型,该模型利用SPSS20.0软件中的主成分分析模块对影响煤层瓦斯涌出量的12个因素进行降维,提取其中3个最能反映原始数据本质特征的主成分因子,再将主成分因子的前25组数据作为训练集,后10组数据作为测试集,借助MATLAB中的LIBSVM工具箱进行支持向量机预测,最后将PCA-SVM、SVM及使用较为广泛的多元线性回归3种方法的瓦斯涌出量预测结果进行对比,预测结果表明PCA-SVM模型在预测精度、稳定性方面都优于其他两种预测方法,更适合煤层瓦斯涌出量的预测。For coal seam gas emission is affected by many factors and shows a complex nonlinear characteristics,based on the theory of PCA and support vector machine(SVM),the prediction model of PCA-SVM for gas emission in coal seam is established.This model uses the principal component analysis module in SPSS20.0 software to reduce the dimensionality of 12 factors affecting the gas emission amount of coal seam,and extracts three principal component factors which can best reflect the essential characteristics of the original data.The first 25 sets of data of the principal component factors are taken as the training set,the last 10 sets of data are taken as the test set and the support vector machine prediction is carried out with the help of the LIBSVM toolbox in MATLAB.Finally,the prediction results of gas emission by PCA-SVM,SVM and the widely used multiple linear regression method are compared.The prediction results show that the PCA-SVM model is better than the other two prediction methods in terms of prediction accuracy and stability,and is more suitable for the prediction of gas emission in coal seam.

关 键 词：煤层瓦斯涌出量 主成分分析 支持向量机 预测 

分 类 号：TD7[矿业工程—矿井通风与安全]