欢迎光临《计算力学学报》官方网站！

Enhancement of genetic algorithm and its application to the identification of soil parameters by inverse analysis[J].计算力学学报,2018,35(2):224~229

Enhancement of genetic algorithm and its application to the identification of soil parameters by inverse analysis
Enhancement of genetic algorithm and its application to the identification of soil parameters by inverse analysis

DOI：10.7511/jslx20170228003

 作者 单位 E-mail 季慧 上海交通大学 土木工程系, 上海 200240 金银富 上海交通大学 土木工程系, 上海 200240南特中央理工大学, 南特 法国 44300 yinfu.jin9019@gmail.com 尹振宇 南特中央理工大学, 南特 法国 44300 吴则祥 南特中央理工大学, 南特 法国 44300 沈水龙 上海交通大学 土木工程系, 上海 200240

本文的主要目的是开发基于实数编码的杂交遗传算法来识别土体的本构参数。该杂交遗传算法在经典遗传算法框架下开发，融合两个新开发的交叉算子，形成了一个新的杂交策略。为了保持种群的多样性，在算法中采用了一个动态随机变异算子。另外，为了提高算法收敛性，采用了一个基于混沌的局部搜索技术。分别基于室内试验和现场试验，通过识别土的本构参数来测试新算法的搜索能力和搜索效率。为了测试新开发算法的突出表现，特选用5种经典的随机类算法（遗传算法、粒子群算法、模拟退火算法、差分算法和蜂巢算法），分析同样的案例进行比较。结果表明，在收敛速度和最优解的准确度方面，新改进的算法可以很好地处理岩土工程的参数反演。

The aim of this paper is to develop a new hybrid real-coded genetic algorithm to identify soil parameters.The new development is under the framework of a classical GA by combining two recently developed and efficient crossover operators with a hybrid strategy.A dynamic random mutation has been incorporated into the new RCGA to maintain the diversity of the population.Additionally,in order to improve the convergence speed,a chaotic local search(CLS) has been adopted.The new GA is applied to identify parameters from an in-situ pressuremeter test and an excavation respectively.In order to highlight the performance of the new GA,5 classic optimization methods(classic genetic algorithm,particle swarm optimization,simulated annealing,differential evolution algorithm and artificial bee colony algorithm) are selected to solve the same problems.The search ability and efficiency of the new hybrid RCGA is estimated by comparisons of all the above methods.