Research on the performance of GMCBO methodology based on model updating of a pedestrian bridge

Consider the loss of diversity of the bodies at the later optimization stage and the low efficiency of the global searching ability of the traditional colliding bodies optimization algorithm in dealing with complex optimization issues of practical structures. Herein, an improved colliding bodies optimization method combing Gaussian-white-noise mutation and innovative collision mode is proposed. The introduction of Gaussian-white-noise is mutating the positions of the selected bodies and enriching the diversity during the whole optimization process. The changed collision mode from 'one to one' to 'one to best' is accelerating the process to global extremum. During searching the global extremums of the 12 test functions which are unimodal, multimodal and multimodal with fix dimension, the proposed shows relatively better accuracy and efficiency. The comparison based on the maximum, minimum, average, standard deviation and non-parametric Wilcoxon test of the optimization results by the above algorithms confirmed the outstanding performance of the proposed method. Through the application of the proposed method to the model updating of a pedestrian bridge, all the differences of the first five frequencies between the measured vibration data and finite element model are reduced from approximate 25% to nearly 5%, and the updated dynamic characteristics show great agreement with the practical structure. From the comparison of the responses between updated model and human-induced vibration experiment, most of the response differences are around 10%. The application of the above real structure further verifies that the proposed method is feasible and strongly recommended in optimization issues of practical structures.