遗传-粒子群算法模型修正

GA-PSO ALGORITHM MODEL UPDATING

用部分测量模态数据对5层钢架结构进行模型修正,将遗传算法、粒子群优化算法、遗传-粒子群组合算法3种算法在该模型修正过程中的效率和精度进行比较,结果表明修正后模型的全部四阶频率和振型都能在不同程度上向目标值靠近,证明3种算法都能够有效修正模型,而且遗传-粒子群算法能在前期利用遗传算法进行高效全局搜索,后期利用粒子群算法进行细致局部搜索,与单独使用遗传算法或粒子群算法相比,组合算法效率和精度更高.

Integrant modal data are used to update a five-layer steel frame. The comparisons between the efficiencies and precisions of the Genetic Algorithm （GA）, Particle Swarm Optimization Algorithm （PSO） and GA-PSO in the model updating processes show that all the four modal frequencies and modal shapes of the updated model can approach the target values with in varying degrees, which proves that these methods can all efficiently update the model. In particular, GA-PSO algorithm uses GA to efficiently search for the globaloptical solution at an early stage, and uses PSO to intensively search for the local-optimal solution at a later stage. Comparing with the PSO and GA, GA-PSO algorithm enjoys higher updating efficiency and precision.