Algorithm Improvement To Dynamic Parameter Identification And Lateral Displacement Monitoring In Structural Health Monitoring System Of Tall Buildings

Super high-rise buildings are emerging in China’s urban construction with the rapid development of China’s economy in recent years.On the other hand,with the development of sensor technology and structural modal parameter identification methods,more structural health monitoring systems have been applied to super high-rise buildings and long-span bridges.Modal parameter identification system has been one of the most important and critical part of structural health monitoring system.In this dissertation,some important models are developed as the supplemental subsystem of existing structural health monitoring systems for Hedong Bridge and Litong Square.Meanwhile improvements of some algorithms are studied in this dissertation on the dynamic parameter identification of super high-rise buildings as well as lateral displacement monitoring technique for super high-rise buildings during construction.The main work in this paper are as follow:(1)The dissertation first analyzed the main programming architecture of the structural health monitoring system on Hedong Bridge and Litong Square.More models such as connection between Lab VIEW and database,time synchronization technology of signal acquisition etc,were further developed as the two major improvement and supplemental subsystems for the original system.Various methodologies to construct the connection between Lab VIEW and database were compared and the method for using Lab SQL tool package to establish the connection with database was studied.Meanwhile key technologies on time synchronization between multiple data acquisition controllers was studied.(2)Some improvements were made for the algorithm of Multiple Random Decrement Technique(MRDT)to be more suitable for the identification of dynamic modal parameters of super tall building with low natural frequency.After clarifying the inherent nature of random decrement method,the random decrement signature is obtained by increasing the length of vibration signal and intercepting the random decrement signature.In this way the obtained signature can have the feature similar to free vibration signal before the following second round of random decrement method.Then more rounds of random decrement method are implemented with different suitable triggering levels and signal length.With the measured wind-induced acceleration signal during a typhoon,the modal parameter identification of Litong Square was conducted by the improved multiple random decrement method.Moreover,the identified results were compared with those identified by stochastic subspace method.It is shown that the improved multiple random decrement method can not only identify the dynamic modal parameters with low-frequencies exactly but also work more efficiently than stochastic subspace identification method.(3)An improved algorithm for conventional Ensemble Empirical Mode Decomposition(EEMD)method was proposed.The improvement includes adding a pair of white noise signals before processing,removing the end effect of empirical mode decomposition by the selfadaptive extending method,deleting Intrinsic Mode Functions(IMFs)whose average frequency does not meet the requirements by clustering analysis to eliminate modal mixture,selecting the main intrinsic mode functions and reconstructing signal by comprehensively evaluating the energy of each IMF and the correlation coefficient between each IMF component and the original signal.An example was selected to prove that the proposed improvement in this paper can effectively reduce the number of stacking,alleviate the end effect and eliminate modal mixture.On the other hand,some improvements were made to deal with the stability diagram of Stochastic Subspace Identification(SSI)method.Therefore the difficulty in selecting the suitable order number for original signal is removed in SSI algorithm and stability diagram can be obtained directly.The pseudo modes in SSI were then filtered by inter-comparing the natural frequencies and damping ratios,the phase angle of each modal vector and the modal vector among all modal points.Finally the correct natural frequency,damping ratio and modal vector for each vibration mode are obtained.With the measured wind-induced acceleration signal of Litong Square during a typhoon,the improved EEMD-SSI method proposed in this dissertation was adopted to identify the dynamic modal parameters of this super tall building.The identified results were also compared with those results obtained from SSI,RDT and power spectrum method.Compared results indicates that the improved EEMD-SSI algorithm can remove the environmental noise and filter the pseudo modes effectively.(4)The improvement of monitoring method on lateral displacement of super high-rise buildings during construction was also studied in this paper.Theoretical analysis and implemental methodologies for precise monitoring of lateral displacement of high-rise buildings during construction were proposed when the total stations for displacement surveying is located in instability area.By constructing the “active” surveying control network with the combined total station and GNSS instruments,the least square method was adopted for precise lateral displacement monitoring when the constructed tall building is actually in a dynamic moving state under the external loading effect.Meanwhile the method for minimizing the number of GNSS control points was proposed by increasing the number of common surveying points for total station.The adjusted parameters and the optimal solution is then obtained by the least square method by constructing observation equations for all the common surveying and control points in the displacement surveying network.By integrating the measured data from biaxial clinometer,the survey accuracy for monitoring the lateral displacement induced by construction of tall building can be reach millimeter level by filtering out the overall lateral displacement caused by a variety of other external loadings.