Research On The Static And Dynamic Characteristics Of Tensile Structure Based On Field Measurement

Large-span tensile structure is an important symbol to mark the level of development of national building science and technology,which exhibits the characteristics of various forms,lightweight and high strength,and spacious building space.Among all the large-span space structures,the cable-net structure is commonly used in the domain of significant public infrastructures,which often have large spans and complex forces.It is highly innovative and challenging in the aspect of design and construction.With the development of the social economy,the increase of engineering practice,and the accumulation of structural service time,the requirements for the construction standards and safety of cable net structures are also increasing,and the problem of structural health diagnosis and treatment has gradually attracted attention from many parties.Therefore,to ensure the safe and stable operation of major projects,this paper focuses on the scientific problem of tracking the static and dynamic characteristics of cable-net structures.Based on the development of static and dynamic measurement technology,three aspects of the research have been carried out: the practice method of monitoring data preprocessing,the wholeprocess displacement monitoring method,and structural dynamic characteristics based on field measurement:(1)Development of static and dynamic measurement technology for tensile structure: This paper proposes a large-area displacement measurement method based on the principle of laser ranging technology,which solves the difficulties of large-area pipeline laying and installation of measurement points in the early stage of construction.It also overcomes the limitation that the displacement of key components cannot be measured synchronously during the construction of traditional large-area space structures.And,this paper also focuses on the Micro-Electro-Mechanical System(MEMS)technology-based field measurement method of large-area vibration response and then proposes a MEMSbased acceleration selection and testing approach,which solves the problem of measurement bias problem during long-term vibration monitoring due to structural deformation or equipment installation deviation.It can provide effective support for structural mode capture.Wireless sensing nodes suitable for static and dynamic monitoring have been developed for different application scenarios.And according to the characteristics of static displacement and dynamic response sampling rate and coverage,the wireless dedicated communication network based on LORA technology and 4G+LORA combined communication is respectively combined to achieve large-area signal coverage and data transmission,meet the monitoring needs of large-scale roofs,which improves the wireless monitoring system.(2)Fault diagnosis technology for abnormal vibration monitoring data: This paper studies the fault diagnosis of abnormal vibration monitoring data in structural health monitoring,and proposes a semi-supervised learning-based fault vibration data diagnosis method.This anomaly diagnosis framework includes two parts: unsupervised learningbased fault data sample classification and supervised learning-based fast batch diagnosis.The wavelet scattering is introduced to extract the fault features of the vibration monitoring samples,and the dynamic time warping algorithm is used to realize the automatic clustering of the sample features and mark the fault sample categories with data labels.Then,the longshort-term memory neural network is used to perform supervised learning on these labeled sample sets,and the neural network model of vibration monitoring data feature classification is obtained,which improves the low effectiveness problem of fault diagnosis of the dynamic time warping algorithm based unsupervised learning.Finally,the practicability and effectiveness of the proposed method are verified by the measured vibration monitoring data on the National Speed Skating Oval(NSSO).(3)Research on the time synchronization method of vibration monitoring data: A data-driven time synchronization approach for measurement points of vibration monitoring is proposed to solve the time synchronization problem in the application of wireless sensor networks.This approach is based on the frequency domain decomposition method in the output-only identification,which combines the frequency-squeezing processing technique to improve the readability of the first-order singular value spectrum.The asynchronous structural mode can be accurately extracted.After that,the corresponding relationship between the inter-channel delay between the measurement points and the phase deviation of the asynchronous structural mode is established,and the optimal estimation of the measurement point delay is finally determined.The accuracy of the proposed approach is verified by three examples: numerical simulation of a four-layer frame excited by white noise,an experimental study of a five-layer steel frame on a shaking table test,and the field measured response of the NSSO.The results show that the proposed approach can correct the time offset between wireless accelerometers as a supplement to hardware synchronization.(4)Research on static displacement monitoring and early warning method for cable network structure: The project overview,structural system,monitoring system,and main construction process of NSSO,which is currently the world’s largest cable-net structure,are introduced.According to the construction process and structural characteristics of the cable-net structure,the construction simulation is carried out to establish the monitoring target of the cable-net structure.Combined with the laser displacement measurement technology,the corresponding static displacement tracking method is proposed.By comparing the simulation results and the field measurement results,the effectiveness of the wireless displacement monitoring is verified to guarantee safety in the construction process,and the experience in displacement control in the construction of this type of structure is accumulated.In addition,during the in-service stage,a static displacement early warning method based on empirical mode decomposition(EMD)and long short-term memory recurrent neural network(LSTM)is proposed.The method includes displacement prediction and threshold estimation.Firstly,the historical displacement time-series signal is decomposed into a series of subsequences by EMD,and the LSTM training models of the corresponding subsequences are constructed respectively.The prediction effect of this prediction model is verified by the simulated signal and the measured signal.Compared with the conventional regression model,this early warning method has better prediction accuracy.In addition,a displacement early warning index DTI based on the Anderson-Darling test statistic is constructed,which overcomes the large subjective selection error of threshold estimation in the traditional peak over thresholdmethod and realizes the optimal displacement threshold estimation.(5)Research on dynamic modal parameter tracking for cable-net structure: Based on the matrix perturbation method,the dynamic modal analysis of the cable-net structure considering the prestress deviation is analyzed.The results show that the cable-net structure is prone to modal jumping.Then an automatic modal identification method of closed frequency structures is proposed,which overcomes the subjectivity and uncertainty of the manual selection by professionals during long-term operation.This method is successfully applied to the dynamic monitoring of NSSO.Through long-term monitoring and tracking,the variation law of the modal damping ratio and modal frequency of NSSO was analyzed,and the correlation between the modal parameters,ambient temperature,and cable force was studied.The evolution law of modal parameters is summarized...