摘要
针对现阶段围岩分级方法存在的主要问题,提出隧道围岩分级的遗传-支持向量分类方法。结合佛岭隧道施工期围岩分级实践,以公路隧道设计规范BQ分级为基准,分别采用岩石回弹强度和掌子面状态替代饱和单轴抗压强度和地应力状态,并增加节理延展性观察的定性指标,在大量现场测试和室内试验的基础上,给出每个分级指标的现场快速测试方法,并以分级结果作为遗传-支持向量分类算法的训练样本,建立隧道围岩分级的遗传-支持向量分类智能模型。佛岭隧道围岩分级实例表明:该模型分级结果与现场勘测基本一致,且较遗传-神经网络模型有更高的分级准确性,为隧道围岩分级提供一种方法。
In view of shortages of the present tunnel surrounding rock classification methods, the genetic-sup- port vector classification method was proposed. The new method was based on the national standard BQ classi- fication system and the surrounding rock classification practice in construction of the Foling Tunnel. The re- bound strength of rock and stability condition of working face were used to replace the saturated uniaxial com- pression strength and crustal stress condition respectively and the qualitative index was added to explore the ex- tension state of joints. Fast testing techniques were given for every classification index on the basis of a large a- mount of field and laboratory tests. The classification results served as the training samples of the genetic-sup- port vector classification algorithm. The genetic-support vector classification intelligent model was established. Application of this model to the Foling Tunnel shows that the results of this model basically agree with site survey work and higher grading accuracy was provided as compared with the GA-BP model.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2013年第1期108-114,共7页
Journal of the China Railway Society
基金
国家自然科学基金(71171016)
交通部部科技项目(2009353334400)
关键词
隧道工程
围岩分级
人工智能
遗传-支持向量分类算法
稳定性分析
tunnel engineering
surrounding rock grading
artificial intelligence
genetic-support vector classifi-cation algorithm
stability analysis
