Automatic Identification Of Structural Modal Parameters Based On Density Clustering

As the basis of structural dynamic analysis,the structural modal parameters have important applications in structural design verification,damage identification and safety assessment,etc.,and modal parameters identification is always one of the research hotspot in the field of structural health monitoring.The stochastic subspace identification(SSI)method has been used widely due to its several advantages,but it requires a lot of manual participation in the identification process,which not only makes the subjectivity and possible difference of the identification results,but also takes time and effort,and is not conducive to online real-time structural damage detection and safety assessment.This paper proposes an automatic modal parameter identification algorithm based on density clustering,and the main research works are as follows:Firstly,the basic theories of linear state space model,covariance-driven stochastic subspace identification and stability diagram,etc.are introduced and the reasons for the occurrence of spurious modes are analyzed.Two numerical simulation examples are used to demonstrate the typical stability diagram shape and modal parameters identification process,and it is pointed out that there are a large number of spurious modes in the stability diagram.To solve this problem,this paper presents a two-stage method to eliminate spurious modes.In the first stage,the soft standard and the hard standard are used to distinguish the true and false modes.In the second stage,the false modes are eliminated based on the stability of the physical modes.This method can eliminate most of the spurious modes,which lays the foundation for the automatic identification of modal parameters.Secondly,a new modal distance is defined to measure the similarity between modes,which can effectively distinguish different modes;the basic theory of three density clustering algorithms is introduced,and the three algorithms are compared and analyzed through two numerical simulation examples.The results show that the three automatic identification algorithms can not only identify the modal parameters of the structure better,but also have high consistency.In terms of identification effect,it is also significantly better than manual identification.A new clustering algorithm is proposed,which does not need to cluster all points.The theory is clearer and the calculation process is simpler.Compared with the other two algorithms,the calculation efficiency is also improved.Finally,three automatic identification algorithms are applied to three practical engineering structures: ASCE’s Benchmark frame model,a cable-stayed bridge and a masonry structure under earthquake aftershocks,further verifying the accuracy,robustness and practicability of the three algorithms.The identification results show that the above method can ensure good identification effect under the conditions of additional noise in the sensor,partial sensor failure,and external excitation completely not satisfying the white noise assumption.At the same time,there is no need to adjust the automatic identification algorithm during the identification process,which proves the universality of the algorithm and realizes the full automation of modal parameters identification.