A type of damage that can lead to catastrophic failure when grows, is crack. Therefore early detection and localization of cracks in structures, is an important issue. In this research a new method for crack identification in structures have been presented. The structure used in this study is a cantilever beam with rectangular cross section. The process of crack identification in the suggested method consists of four steps. In first step, Using finite element method first three natural frequencies of the structure for different locations and depths of cracks were obtained. In second step, a feed forward neural network was created. In third step, a computer code based on genetic algorithm optimization method were defined and used to training the neural network. Also, a back error propagation neural network was trained. Inputs of neural networks were first three natural frequencies and outputs were locations and depths of cracks. In forth step, natural frequencies of some beam with distinct crack conditions as inputs applied to trained neural networks and obtained results from two methods were compared with each other. Obtained results were shown that cracks characteristics were computed with good approximations.

Keywords