Connection Loosening Detection Of Frame Supported Glass Curtain Walls Based On Coupled Structural Response Vector And Support Vector Machine Approach

Architectural glass curtain wall has become an increasingly indispensable part in modern buildings.While in the actual use,there may be risks of connections loosening or even falling off,which seriously affects the safety of public use.To date,there is still no effective and practical monitoring approach for real-time detection of connection loosening condition of glass curtain walls.The most common way for safety evaluation is to carry out simple visual inspection of building glass curtain walls based on relevant specifications or destructive mechanical performance experiments on a small number of glass curtain walls.Therefore,it is of great necessity and significance to propose an effective and reliable identification method for the connection loosening of frame supported glass curtain walls.A coupled structural response vector and support vector machine approach(SRVSVM)was developed for the connection loosening detection of frame supported glass curtain walls based on theoretical research and numerical simulation verification.The main contents are as follows:Firstly,the development of glass curtain wall was reviewed and the significance and main research background of the research were revealled.The literature research and measurement analysis method were used to analyze the core pain points of the loose identification of the frame supported glass curtain wall.The frequent incurrence of current glass curtain wall failure accidents and lack of practical existing monitoring methods were found.In the meanwhile,the rapid development of damage identification theory and sensor technology may also accelerate the realization of the connection loosening identification of the frame supported glass curtain.Secondly,the connection loosening of frame supported glass curtain wall was examined.Based on the results of comparative experiments,a theoretical analysis model for the connection loosening detection of frame supported glass curtain walls along with its finite element model were proposed.The validity of the theoretical model and the finite element model was confirmed by comparing with the experimental values of the center measuring point displacement and the calculated values in the finite element.A sample dataset for connection loosening detection was established using python.Thirdly,the theory of structural response vector(SRV)and its application in structural damage identification were summarized.The structural response vector(SRV)with dynamic and static components was used to identify the connection loosening mode and levels.The structural response vector was then used for damage pattern identification and quantification,which validated the effectiveness and applicability of SRV as damage indicators.Then,the basic principles of support vector machine(SVM)for structural damage identification were reviewed and a coupled structural response vector and support vector machine approach(SRV-SVM)for frame supported glass curtain wall was developed.Different SRVs with the first four frequencies of glass panels and the inclination angles at nine points were used as the input of the support vector machine(SVM).Three SVMs with different kernels functions were used for the sample training and damaged identification.Results showed that the satisfactory effectiveness of coupled SRV-SVM approach for the connection damage identification of frame glass curtain walls.Finally,in order to facilitate the support vector machine(SVM)kernel function selection and to propose the parameter optimization strategy,a particle swarm optimization(PSO)was used to optimize the SRV-SVM approach proposed and the identification results of optimized SRV-SVM and other three SVMs were compared.Results showed that the optimized SRV-SVM approach has a higher accuracy and effectively decreases the numbers of sensors in the frame supported glass curtain wall connection loosening detection.