基于高斯过程机器学习的岩爆等级识别方法  

Method of Identifying Rockburst Grades Based on Gaussian Process Machine Learning

作　　者：张研[1] 苏国韶[1] 燕柳斌[1] 

机构地区：[1]广西大学土木建筑工程学院,南宁530004

出　　处：《地下空间与工程学报》2011年第2期392-397,共6页Chinese Journal of Underground Space and Engineering

基　　金：国家自然科学基金资助项目(50809017);中国博士后科学基金及特别资助(20080440812;200902354)

摘　　要：针对岩爆发生等级与其影响因素之间存在着复杂的非线性关系,如何根据影响因素有效识别岩爆等级是一类复杂的模式识别问题。提出了一种基于高斯过程二元分类模型的岩爆等级识别方法,该方法通过对少量学习样本的学习,就可以建立岩爆等级与其影响因素之间的复杂非线性映射关系。将方法应用于锦屏二级水电站长探洞和引水隧洞岩爆等级实例,研究结果表明,该方法具有模型参数自适应确定、容易实现且识别精度高等优点,具有良好的工程应用前景。Aiming at the fact that it is still difficult to reasonably identify the rockburst grades,the method based on GPC model is proposed for identifying rockburst grades.According to a few learning samples,the nonlinear mapping relationship between rockburst grades and its influencing factors is established by GPC model.The method is applied to the long exploratory tunnel and diversion tunnel of Jinping secondary hydropower station.The results of real engineering study show that the method is feasible,simple to be implemented and precise,that makes itself very attractive for a wide application in identifying rockburst grades.

关 键 词：岩爆等级 识别 高斯过程 机器学习 

分 类 号：TU457[建筑科学—岩土工程]