| The spatial structures,represented by the National Stadium(the Bird Nest)and Tianjin National Convention and Exhibition Center,play an important part of modern architectures.These buildings not only have their own functionality,but also have great economic,cultural and political significance.Structural health monitoring(SHM)technology is an important means to ensure the structural safety of building structures during the construction and operation stages.Due to the characteristics of spatial structures,SHM technology for such structures is usually unapplicable.Therefore,it is of great practical significance to carry out SHM research on spatial structures.For the SHM of spatial structures,the relevant research shows that the commonly used damage indexes in the data-driven damage identification methods often fail to be obtained through the in-situ tests,and the data analysis methods have high requirements for accuracy and robustness.Based on the above reasons,this thesis proposed a damage identification framework especially for spatial structures based on a novel LALAda Boost algorithm,including data preprocessing and anomaly detection,damage features extraction and damage analysis.Firstly,a data anomaly detection method based on LAL-Ada Boost algorithm was proposed,aiming at detecting five common data anomaly patterns in SHM scenarios,and was verified by dataset collected from a grid experiment.The results showed that the proposed algorithm could perform excellently with high accuracies and fast converging speed.Secondly,a damage index based on frequency-domain analysis considering accelerometers displacement was proposed as the input of machine learning algorithm;ANSYS was used for finite element modeling of the grid experimental structure,and six different damaged scenarios were simulated.Using the above damage indexes as the input for LAL-Ada Boost algorithm,four monitoring scenarios are verified respectively: single damage degrees detection;minor damage degree detection;multiple damage degrees detection;damage location.The results turned out to be accurate for each monitoring scenario.Last,the robustness of the algorithm was verified through anti-noise tests.The proposed framework was then further verified on a SHM project of Tianjin No.1 High School Stadium and the results showed the feasibility of the approach in this paper. |
