基于遗传退火混合算法的弹性地基上框架结构参数识别研究

对传统的简单遗传算法(GA)进行了改进,融合模拟退火技术(SA)的思想,建立了遗传模拟退火算法(GASA)的串行结构.GA采用群体并行搜索,通过概率意义下基于"优胜劣汰"思想的群体遗传操作来实现优化.SA采用串行优化结构,赋予搜索过程一种时变最终趋于零的概率突跳性,避免局部极小并最终趋于全局最优.两者的结合提高了遗传算法的全局搜索能力.本文对一实验室中弹性地基上框架结构进行了逐层模态实验研究,得到了四种工况下的模态频率和振型.首先对利用GASA算法对退火参数进行了优选,SA部分中的退温参数g和扰动幅度参数η对搜索效率及全局搜索能力具有重要的影响;然后对四种工况下混凝土的弹性模量和地基的动剪模量进行了识别,并与灵敏方法识别结果进行了对比,得到了结构物理参数随着结构浇注层数的增加而上升的规律,识别得到的弹性模量比回弹法结果偏大,与结构的静模量和动模量的区别有关.以上方法及其应用对于结构的健康监控具有现实的意义.

Parameter identification of the frame structure on elastic foundation based on the cenetic annealing hybrid algorithm

In this paper the simple genetic algorithm is improved by syncretizing the ideology of annealing algorithm and the genetic annealing algorithm.GA adopts population parallel search,it based on the idea of 'fitness is survival' to come into optimization.SA adopts serial structure,the process is endowed with time-variety probability abrupt jumping character,the local limit is prevented and the global optimization is achieved.and the globe searching ability is developed by combining the two methods.The modal te...