摘要
岩体力学参数对地下洞室的施工和设计是极其重要的,通过位移反分析确定岩体参数的应用研究很多。为了提高参数反演的效率,采用均匀设计理论设计不同的反演参数组合,用有限元计算组建的样本集训练神经网络,根据洞室开挖过程中实测位移反演地下洞室岩体力学参数。基于面向对象、自适应、可视化三大关键技术,开发了地下洞室岩体力学参数反演有限元软件系统BAFES。该软件可实现地下洞室力学参数反演分析、开挖预报过程的可视化,将地下洞室岩体力学参数反演模式由“数据-有限元分析-数据”转变为“图形-有限元分析-图形”,结合现场监控技术和工程稳定分析,能及时地指导工程设计和施工。以某水电站地下厂房洞室群岩体力学参数反演为例,说明了该软件的实用性。
Rock-mass parameters are very important to underground cavern design and construction; many application studies about calculating rock-mass parameters have been implemented by back analysis of displacements. In order to improve parameters back analysis, the uniformity design theory is introduced to design training swatch for neural networks, rock-mass mechanical parameters of underground cavern are obtained from measured displacement messages. The back analysis finite element software is implemented based on three key techniques： object-oriented, adaptability and visualization, which can be used as visualized back analysis and forecast tool. Rock-mass parameters back analysis is implemented by mean of ＂graph--finite element analysis--graph＂ instead of ＂data--finite element analysis--data＂, it can instruct design and construction in time with monitoring technique and stabilization analysis. It is illustrated that this software is very useful bythe example of the underground cavern of a hydropower project.
出处
《岩土力学》
EI
CAS
CSCD
北大核心
2007年第3期577-581,共5页
Rock and Soil Mechanics
基金
国家自然科学基金(No.50239070
No.50379039)
关键词
地下洞室
岩体参数
反演分析
面向对象技术
神经网络
有限元分析
underground cavern
rock-mass parameter
back analysis
object-oriented technique
neural networks
finite element analysis
