摘要
传统的信息化设计在监测信息处理、信息的反馈分析及初步设计的信息化修正等方面存在不准确、不及时和难以操作的缺点。本文结合铜黄高速公路坞石隧道的施工,将一种改进的支持向量回归(SVR)算法引入隧道工程的信息化设计中。采用十进制遗传算法搜索改进的SVR参数,形成改进的GA-SVR算法,并编制相应的计算程序。坞石隧道的应用结果表明,这种改进的GA-SVR算法无论是对监测数据的拟合预测,还是监测信息的反馈分析都能做到准确快速,在反分析完成后采用改进的GA-SVR算法进行3个开挖步内的位移超前预报也具有较高的精度。最后提出一种基于此算法的初步设计最优化修正方法,形成以此算法为核心的完整的隧道工程信息化设计方法,具有快速、操作简单和计算准确的优点。该算法可以在隧道工程中使用。
The conventional information design method has the disadvantages of being inaccurate, slow and inconvenient in monitoring data processing, information feedback analysis and initial design correction. In this paper,the improved support vector regression(SVR) algorithm is introduced into the field of tunnel engineering information design combined with the construction of the WUSHI Tunnel on the Tongling-Huangshan High- way. The decimal genetic algorithm is used to search for the parameters of the improved SVR model to form the improved GA-SVR algorithm, and the corresponding source codes are programmed. The application results in the WUSHI Tunnel show that the improved GA-SVR algorithm can reach high precision and speed in monito- ring data fitting and forecasting and in monitoring data feedback analysis and the improved GA-SVR algorithm can also acquire a better displacement prediction accuracy within three construction steps after displacement back analysis has been completed. Finally, the initial design optimization method based on the improved GA- SVR algorithm is brought forward to form the complete tunnel engineering information design system with the improved GA-SVR algorithm as the core. The proposed system is fast in processing, simple in operation and accurate in calculation. It can be applied in urban Metro and railway tunnel engineering construction widely.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2008年第4期71-78,共8页
Journal of the China Railway Society
基金
国家自然科学基金(50078002)
关键词
隧道工程
支持向量回归
遗传算法
位移预报
信息化设计
tunnel engineering
support vector regression
genetic algorithm
displacement prediction
informarion design