基于改进蜂群算法的结构损伤识别

Damage identification of coupled system based on modified ABC algorithm

人工蜂群算法是一种元启发式算法,具有构架简单、易于操作、鲁棒性较好的特点。本文对原始蜂群算法进行了完善:在食物源的遴选方式上,采取锦标赛机制代替标准算法的轮盘赌机制;在跟随蜂阶段引入局部搜索能力更强的公式来刻画蜜蜂采蜜过程。将结构的损伤归结为单元刚度的折损,然后利用频率残差和模态确保准则(MAC)构建问题的目标函数,再使用原始算法和改进的算法求解该非线性优化问题,得到最后识别结果。算例表明,改进算法比原始算法能够更加有效地识别出局部损伤,并且抗噪能力更强。

Artificial bee colony（ABC） algorithm is a heuristic algorithm with simple structure and ease of implementation and robustness. In order to make it more powerful, some improvements are presented： the roulette selection strategy is replaced by the tournament selection mechanism and a new formula is used to simulate the onlooker bee＇s behavior to enhance exploitation ability. The objective function for the damage identification is built based on frequencies and MAC. Then, the ABC and modified ABC（M-ABC） algorithms are used to solve this nonlinear optimal problem and thus acquires identified results. Final results show the proposed technique produces better parameter estimation, even with noise corruption, comparing with the original one.