Research On Structural Damage Identification Method Based On Time Series Model

Structural health monitoring(SHM)is critical to assessing the condition of engineering structures and ensuring its safeties.In order to build a practical and efficient SHM system.One of the key tasks is to rapidly improve its ability to identify structural damage.Damage identification methods based on vibration information have been widely studied for decades,and the literature in this area is quite abundant.However,due to complex and changing operating conditions and environmental conditions,as well as various and complicated actual engineering structure forms and damage types,the current damage identification methods still face huge challenges in practical applications.The time-series analysis-based damage identification method is an important member in the field of vibration information damage identification.Currently,there is still much room for exploration in such issues as damage localization,online damage monitoring,large-scale structural damage identification,and nonlinear damage identification.This research proposes a novel structural damage indicator based on the combination of autoregressive(AR)model and multi-target tracking.It is hoped to provide reference ideas for improving the shortcomings of current identification methods.This article first reviews the current development and problems of damage detection methods,and then proposes the main research ideas of this paper.First,this study proposes a new type of structural damage indicator,which is defined as the optimal subpattern assignment distance(OSPA)between AR model poles.The solving of the pole–based OSPA distance is accomplished by Hungarian algorithm.Compared with the structural frequency indicator that reflects the overall damage of the structure,the pole-based OSPA distance improves the local damage localization ability.Compared with the AR model residual indicator(RVD),the polebased OSPA distance has a better damage quantification ability.The numerical simulation and experimental analysis of damage identification is conducted on a five-story shear building structure,which shows a clear improved result.Secondly,in order to explore the possibility of embedding the OSPA indicator into the design of the SHM system,this study simulates the online damage identification based on continuous and same time intervals of the OSPA indicator.Since the actual engineering structure is often a complex three-dimensional structure,structural responses of different positions and different orientation need to be measured in order to assess its structural damage more fully.At present,although there are many structural damage indicators,few have the ability to fuse multi-sensor information,while the OSPA indicator itself has the ability to fuse multi-target features.Here,this study explores the structure damage identification on IASC–ASCE benchmark structure using the proposed pole-based OSPA indicator.The results show that the OSPA indicator can effectively locate single-floor and multifloor damages of this three-dimensional structure,and also have satisfactory damage quantification.Next,in response to the challenges that large-span structures bring to the identification accuracy of previous methods,this study proposes a substructural damage identification method based on the autoregressive moving average with input(ARMAX)model.The original damage indicator is extended to be based on an improved version using the ARMAX model.In the experimental analysis of the damage identification on a five-story shear frame structure provided by the Building Research Institute of Japan,the identification of RVD indicators without substructure division is not effective,but the RVD indicators of ARMAX models based on the divided substructures have improved significantly.Thus,the effectiveness of the substructural detection method is verified.Then,the damage identification of the 12-story reinforced concrete frame structure from Tongji University was carried out.The result of the OSPA indicator based on the AR model poles was far from the damage phenomenon after the shaking table test.After the substructure division,the results of pole-based OSPA distance between ARMAX models are more in line with the actual damage phenomenon.Therefore,the proposed substructural damage detection method can provide guidance for the current detection method to improve the recognition ability of actual large-scale structures.In addition,there are many non-linear damages in actual engineering structures,such as breathing cracks during vibration of components.Nowadays,the research on non-linear damage indicators is much less than the linear damage indicators,and the linear damage indicators often have limited ability to identify non-linear structural damage.This paper introduces the autoregressive conditional heteroscedasticity(ARCH)model in the financial field,and constructs a damage indicator called CVD,which is based on the change of conditional variance.The detection results with the linear RVD indicator and OSPA indicator results are compared with the ones of CVD.Moreover,two improved non-linear damage identification methods are further proposed,namely AR-GJR(jointly proposed by Glosten,Jagannathan and Runkle)model based method,and ARMAX-ARCH model based method.Three nonlinear structures are used to evaluate the performance of proposed non-linear damage identification methods,and they are the nonlinear breathing crack beam model offered by Prof.Kullaa,Alamos National Laboratory’s three-story nonlinear shear building structure and eight-degree-of-freedom nonlinear dynamic system.It shows significant improvement over the previous methods.Finally,the end of this paper summarizes the main research content and provides valuable future research prospects that have not yet been conducted.