Comparison of different global optimization algorithms for model updating with application to a full-scale bridge structure

Due to inherent simplifying assumptions in the finite element (FE) models, the actual behaviour of full scale structures often differs resulting in incorrect detection of the dynamic response under seismic conditions. On site measurements may reveal important differences between measured data and predictions from an FE model. In model updating, dynamic measurements such as natural frequencies, mode shapes and damping ratios are correlated with their FE model counterparts to calibrate the FE model. In this paper, different optimization techniques for model updating have been investigated. Different global optimization algorithms (GOAs), including particle swarm optimization (PSO), genetic algorithms (GAs) and simulated annealing (SA), were used for model updating. The results are compared in terms of accuracy of the updated frequencies. The first part of the paper gives the details of the modal testing of a full scale cable stayed footbridge. The bridge is composed of two spans with composite steel concrete deck, a centrally located steel pylon and six pairs of stays. The bridge was excited using three dynamically synchronized shakers. A dense array of sensors was employed to measure the response. The second part describes model updating of the bridge FE model. The aforementioned GOAs were used to calibrate the FE model with the experimental results. The paper concludes with a discussion on the efficacy of using the different GOAs to obtain a representative FE model.