基于遗传算法的混合优化反分析及比较研究

COMPARISON STUDY OF MIXED OPTIMAL METHODS BASED ON GENETIC ALGORITHM IN BACK ANALYSIS

围绕优化反演分析中计算收敛速度、精度和稳定性问题,着重就传统和现代两类优化方法开展以下3方面的研究:(1) 将阻尼最小二乘法与遗传算法耦合起来,发展了阻尼最小二乘法-遗传混合算法;(2) 将两类混合优化方法:阻尼最小二乘法-遗传算法和模拟退火-遗传算法较早地用于优化反演分析;(3) 结合基坑工程算例,对单纯形法、阻尼最小二乘法、遗传算法、模拟退火-遗传算法和阻尼最小二乘法-遗传算法进行了比较分析。结果表明,与单纯形法等传统的优化方法相比,基于遗传算法的一类现代优化方法具有较好的全局收敛性;与常规的遗传算法相比,阻尼最小二乘法-遗传和模拟退火-遗传等算法有效地提高了优化反演的计算搜索速度和精度。

In light of computing speed,precision and reliability,the following three problems related to traditional and modern optimal methods are mainly studied: (1) the mixed genetic algorithm is presented based on the idea of coupling genetic algorithm with damping least square technique;(2) the above mixed genetic algorithm and simulated annealing-genetic algorithm are led into back analysis in geotechnical engineering early;(3) some optimal methods,such as the simplex method,damped least square technique,genetic algorithm,simulated annealing-genetic algorithm and the mixed genetic algorithm,are evaluated and compared. The studied results show that compared with the traditional optimal methods like the simplex,the modern one based on genetic algorithm is of better convergence in whole region,and compared with the common genetic algorithm,the mixed genetic algorithm and simulated annealing-genetic algorithm effectively improve searching speed,precision and reliability of genetic algorithm.