GA-based Model Updating for Existing Bridges

In this research, a model updating technique based on GA(genetic algorithm) is developed and applied to an existing PS concrete girder bridge. Usually, numerical model updating requires long hours of optimization process and has troubles in accessing to a reliable target solution. To reduce the computational time and improve the accessibility, we developed a program whose whole process runs only in Matlab environment. The parameters used in the proposed model updating process include the design range of the crossover probability, mutation probability, and the step parameter adjusting the amount of increment of the parameter. It was confirmed that the model updating performance is improved when the gene arrangement of the chromosome is closely related to the property of an actual bridge and the static and dynamic characteristics are well applied to the fitness function. The importance of selecting parameter values was determined by complex model formation and the design range of reasonable parameters of the current model.