基于支持向量机的赵官矿小断层预测  

Forecast of small fault based on SVM in Zhaoguan mine

作　　者：谭希鹏[1] 施龙青[1] 邱梅[1] 徐东晶[1] 季小凯 王娟[1] 

机构地区：[1]山东省沉积成矿作用与沉积矿产重点实验室山东科技大学地质科学与工程学院,山东青岛266590

出　　处：《煤田地质与勘探》2015年第5期11-14,共4页Coal Geology & Exploration

基　　金：教育部高等学校博士学科点专项科研基金项目(20133718110004);青岛经济技术开发区重点科技发展计划项目(2013-1-62);山东省自然科学基金重点项目(ZR2011EEZ002);山东科技大学科研创新团队支持计划项目(2012KYTD101)

摘　　要：针对支持向量机(SVM)适合小样本数据建模这一特点,提出利用SVM进行煤矿小断层预测。以小断层广泛发育的赵官矿实测数据为基础,选取断层倾向、落差、倾角以及断层性质作为特征影响因子,以Matlab中的libsvm工具箱为平台,总结了SVM建模过程,并建立了赵官矿井小断层水平延展长度的SVM预测模型,验证了该模型在该井田内具有一定泛化性。为验证SVM小样本建模的优越性及该模型在相似地质条件下工程实例的应用,利用该模型及传统的多元回归模型对赵官矿以及邻矿——邱集矿采掘工程中新揭露的小断层进行预测,并将两种方法计算结果与实际结果进行对比,发现SVM预测结果更精确。Because of its fitness for small sample modeling, a model based on SVM was established. The model used for forecasting the horizontal extending length is combined with the actual measured data of 7# coal seam. The tendency, throw and dip angle of fault was the impact actors of the forecasting model. The software package called libsvm based on Matlab was the platform of the model. Taking advantage of the model, three samples in Zhaoguan mine and three samples in Qiuji mine were perfectly predicted. Compared with the result from multiple regression analysis,SVM has a better outcome when the sample is small.

关 键 词：支持向量机 赵官井田 小断层预测 

分 类 号：P542.3[天文地球—构造地质学]