基于粒子群优化算法的结构模型修改

结构模型修改已经演化为一个多学科的研究课题．在最优化框架内，应用了国际上最近提出的粒子群优化算法，该算法具有全局搜索能力并且不需要目标函数的解析表达式。对于一实际钢结构，利用部分和全部测量得到的模态数据进行了模型修改的实验研究．并与基于灵敏度分析、神经网络和遗传算法的模型修改方法进行了对比．以修改后模型计算出的模态数据与实验测得的模态数据的相似度来衡量模型修改的准确性。结果表明，在多数情况下，所提出的模型修改方法得到了最好的修改结果，因此，应用粒子群优化算法进行结构模型修改是可行的。

Structural Model Updating Based on Particle Swarm Optimization

Structural model updating has been evolved into a multidisciplinary research subject. In this paper, the problem is solved in the framework of optimization. Particle swarm optimization(PSO) algorithm is applied, which is proposed recently as inspired by some mechanisms in sociology, psychology, and ecology and has distinguished global search capability and does not need explicit expression of objective functions. For a real steel structure, some model updating experiments are carried out using partial and complete experimentally measured modal data, respectively. The updated results are compared with some methods based on sensitivity analysis, neural network, and genetic algorithm. Precision of model updating is measured by the similarity between experimentally measured modal data and predicted modal data with updated models. Comparisons indicate that the propose model updating method gives the best results in most cases. Soupdating structure model with PSO is valid.