Structural Damage Identification Based On Prior Information And Compressed Sensing

More and more large-scale civil and infranstructure engineering have been constructed since China has entered the new era of modern construction.Structural safety and reliability during service have attracted a lot of attentions.In order to assess the status of structures,structural health monitoring?SHM?systems have been proposed.As one of the main techniques in the SHM,structural damage identification?SDI?has been widely studied.In recent years,several SDI methods have been proposed based on sparse regularization.This type of methods can effectively improve the ill-posedness of the original SDI methods;however,some defections still exist in these methods.In order to rationalize identified results,enhance identified efficiency and reduce computational space,two improved methods which combine existing methods with norm normalization,prior information of damages and compressed sensing,are proposed for SDI.The contents of this dissertation are summarized as follows:?1?Vibration-based SDI approaches proposed in recent years are reviewed.According to whether physical characteristics are used to identify damages,SDI approaches can be categorized as:modal-based method;local diagnostic method;non-parametric method and time series method.The above methods are briefly summarized,and the applicability of introducing sparse regularization and compressed sensing into SDI are also described briefly.?2?An improved SDI method is proposed based on norm normalization and prior information of damages.Due to the influence of noise,misjudgments and stiffness hardening will be observed in the SDI results when the first-order sensitivity analysis method is combined with the sparse regularization.On the basis of this method,norm normalization is introduced to adjust the sensitivity of each element;at the same time,different kinds of prior information of damages are added to restrict the solution space by considering the actual situation of structures.The robustness,validity and applicability of the proposed method have been verified by numerical simulations of cantilever beam,two-story rigid frame and two-dimensional truss.?3?A new SDI method is also proposed based on compressed sensing and prior information of damages.To combine first-order sensitivity analysis with compressed sensing,both a redundant dictionary related to sensitivity matrix and a measurement matrix satisfying Dictionary-restricted isometry property?D-RIP?are first constructed.A subspace projection method is then adopted to compress computational space of damage identification.Finally,the l₁-norm regularization method is used to solve the objective function,the prior information of damages is introduced to obtain reasonable SDI results.The capabilities of the proposed method via compressing computational space and identifying structural damages are studied by the numerical simulations of a two-dimensional truss.?4?A cantilever beam is fabricated in laboratory and used to verify the correctness and validity of the SDI method based on compressed sensing and prior information of damages.The frequencies of beam in healthy status are obtained by the experimental modal analysis,and particle swarm optimization algorithm is introduced to revise some relevant parameters in finite element model so that the model can describe the actual beam well.Several damage cases are simulated by reducing different local width of beam.So structural frequencies in different damage cases are obtained,and the damage statusof beam is assessed by the proposed method.According to SDI results in damage cases,the feasibility of the proposed method is verified well.