摘要
鉴于在基坑工程中传统位移反分析方法存在的计算量较大,不易掌握以及求得岩土介质力学性能参数精度较低的不足,提出基于克隆选择算法的位移反分析方法。该方法的核心在于采用BP网络为正演工具代替有限元方法,描述岩土介质力学性能参数和位移的非线性映射,以减少计算量;利用克隆选择算法为反演工具,优化岩土介质力学性能参数,以提高计算精度;依据高精度的岩土介质力学性能参数又可以估算出高精度的位移。仿真试验表明了该方法的可行性和有效性。
Based on the fact that the conventional displacement back-analysis methods have disadvantages of complicated computing, not being easy to catch, and low precision solutions, this paper proposes a new method to solve the displacement back-analysis problem based on the clonal selection algorithm. The main point of this method is to use BP nets as a tool to replaces plastic FEM program in the forward process and to map the non-linear relation of samples and reduce computed data. The clonal selection algorithm is used as a tool in the backward process to optimize index parameters and improve computing precision, and at the same time, a high prcision displacement can be achived based on the high prcision index parameters. Simulation results show that the proposed method is not only feasible but effective.
出处
《成都理工大学学报(自然科学版)》
CAS
CSCD
北大核心
2009年第2期205-209,共5页
Journal of Chengdu University of Technology: Science & Technology Edition
关键词
基坑
位移反分析
克隆选择算法
BP网络
foundation pit
displacement back-anaysis
BP nets
clonal selection algorithm