基于遗传算法的结构损伤诊断研究

利用试验得到的振动参数评估结构的破损情况，是当前结构工程学科十分活跃的领域。本文引入遗传算法处理振动参数，进行结构的损伤诊断。为了让遗传算法更适用于结构工程损伤诊断领域，提出了多父体变量级杂交和变量微调等新的改进策略，并运用于固端梁的损伤诊断，取得了满足工程要求的结果。在识别过程中，遗传算法用到的信息有前二阶固有频率和其相应的振型及前三阶固有频率和第一阶振型，充分说明了遗传算法可在已知信息不多的情况下找到最优解，为结构损伤诊断开辟了一条新路。

DAMAGE DIAGNOSIS OF STRUCTURES BY GENETIC

A method based on the genetic algorithms is developed for the estimations of structural damage using the measured dynamic data. Some improved strategies, for instance, multi- parents crossover and fine-tuning of variables, are introduced to detect the damage of a fixed-end beam, and satisfactory results are obtained in the paper. In the calculation, two groups of data are used as inputs of the genetic algorithms, that is, the first two order natural frequencies and the corresponding vibration modes and the first three natural frequencies and the first vibration mode. It is concluded that optimum is achieved successfully by using the genetic algorithms with less data.