摘要
将支持向量机与遗传算法相结合,提出了一种用于位移反分析的进化支持向量机方法。这种方法基于试验设计和有限元计算获得学习样本和检验样本,用遗传算法搜索最优的支持向量机参数,用获得的最优模型进行学习,从而获得岩体的力学参数与位移之间的非线性映射关系,再用遗传算法从全局空间上搜索,进行岩体力学参数的识别。给出的两个算例结果是令人满意的。
An evolutionary support vector machine for displacement back analysis is proposed by combining the support vector machine and genetic algorithm. The learning and testing samples produced in orthogonal and equality experiment are used to train the support vector machine whose parameter is determined in global optimal by genetic algorithm. Thus, the support vector machine with optimal parameter is used to describe the relationship between the rock mechanics parameters and displacements. Then genetic algorithm is adopted again to search for the optimal rock mechanics parameters in their global ranges. As an example, a back analysis for elastic and elasto-plastic problem is introduced. The results are satisfactory.
出处
《岩石力学与工程学报》
EI
CAS
CSCD
北大核心
2003年第10期1618-1622,共5页
Chinese Journal of Rock Mechanics and Engineering
基金
中国科学院知识创新重要项目(KJCX2-SW-L1-3)
国家自然科学基金(50179034)
国家重点基础研究发展规划(973)项目(2002CB412708)资助。
关键词
位移反分析
支持向量机
遗传算法
有限元
岩体力学
Elastoplasticity
Finite automata
Finite element method
Genetic algorithms
Global optimization
