Localized Crack Identification in Steel Structures Based on Polynomial Chaos Expansion
Crack prediction is one of the important research areas in the field of structural health monitoring.A polynomial chaos expan-sion(PCE)surrogate model was used to predict the local damage of steel beam structures.The PCE surrogate model employs initial samples to establish the link between the response and the damage parameters in place of the original structural response and its physi-cal parameters,and effectively reduceed the repetitive invocation of the finite element software to delineate the finite element mesh and the time required for finite element computation during damage identification,improving the identification efficiency;in addition,a sparse PCE construction algorithm combining Bregman iteration and greedy coordinate descent method was used.Numerical experiments showed that the PCE model's prediction accuracy and computing efficiency were much higher than the optimisation algorithms optimis-ing artificial neural network models(BOA-ANN,PSO-ANN,GA-ANN).It is proved that the method proposed in this paper shows sig-nificant effects in simply supported beam structures and plate structures,and provides a theoretical method for engineering structural damage identification and assessment.
structural health monitoringcrack predictionpolynomial chaos expansionsurrogate model
NETL
NSTL
万方数据
