Detection of multiple cracks in beams using Genetic Algorithms

The detection and identification of structural damages play a key role in the monitoring and servicing of structural safety and reliability. This paper applies the real-coded Genetic Algorithm to the crack detection problems of multi-cracked beams. The objective function of the Genetic Algorithm was minimized by comparing the changes of measured vibration data of cracked and uncracked structures with those of the analytical model before and after updating. In the analytical model, the massless spring was used to simulate the open crack in the beam. Based on the frequency changes and the mode shape changes caused by the presences of cracks, the analytical model is updated to find the crack locations and sizes. A thorough explanation on how to find the mutation rate, crossover rate, population size, and weights of the vibration data are also discussed.