Online Tracking And Identification For Operational Modal Parameters Of Long-Span High-Speed Railway Bridges

Modal parameters are inherent characteristics of structures,and their accurate identification is important for model modification,damage identification and state assessment.Operational modal tracking and identification technology has become one of the research hotspots in structural health monitoring because it only needs responses to grasp the variation of structural dynamic characteristics.This thesis takes the long-span high-speed railway bridge as the research object.Based on the random vibration responses of bridge before train excitation,the stochastic-deterministic vibration responses of bridge during train excitation,and the short-time free vibration response of bridge after train excitation,the online tracking and identification method is studied,where the problems about spurious modes,mode missing,weak mode identification are solved.Besides,the early warning of bridge abnormality is preliminarily studied through long-term tracking modes.The main contents are as follows:(1)Before train excitation,the random vibration responses of bridge have low signal to noise ratio since the low-energy pseudo-broadband ambient loads are acted on the bridge.The low signal to noise ratio leads to the spurious modes and mode missing in identification results.According to the characteristics that structural modes under different model orders are similar,the similarity indicators to distinguish between structural and spurious modes are defined.The spurious mode elimination method based on the clustering of the similarity indictors is proposed.Besides,the relation between modal observability vectors in various times is illustrated,the complete observability vector spaces are spanned to avoid the missing of identified modes,the tracking method based on the maximum of observability vector similarity is established.As the examples show,the proposed method could track structural modes from random vibration responses with different signal-to-noise ratio and variable vibration bandwidths.(2)During train excitation,the weak modal components of the structure are easily submerged by the excitation components in the stochastic-deterministic vibration response of bridge,which is acted by the high energy narrow-band train load dominantly.To identified the weak modes,the mapping relation between the bridge subsystem vibration and stochastic-deterministic subspace model is constructed,the experimental modal analysis idea is introduced,the virtual vehicle moving load series are constructed based on field monitoring data,the equivalent relation between virtual vehicle moving load series and deterministic loads of bridge nodes is established,the enhanced stochastic subspace identification method based on the projection of response-equivalent load is proposed.Examples show that the proposed method strengthens the proportion of structural modal component to the analytical response,realizes the weak modal identification under finite model orders,and solves the problem that the parametric operational modal analysis methods are not suitable for the analysis of bridge vibration under train excitation.(3)After train excitation,the vibration responses of bridge under high initial condition have no significant feature,which makes it difficult to extract the effective free vibration.This paper compares the characteristics between multiple degree-of-freedom responses and single degree-of-freedom components,reveals that the amplitude feature of the single degree-of-freedom component can be used to separate free vibration and forced vibration,proposes the iterative 2-variational mode decomposition method to decompose the vibration response without mode aliasing,establishes the free vibration extraction method based on the envelope attenuation characteristic of the decomposed component.Examples show that the proposed method ensures the requirement of effective free vibration data for high-precision identification,and realizes online switching of the identification methods suitable for different vibration forms.(4)Considering that the modal parameters can be used as an important index to evaluate the structural dynamic performance,this paper preliminarily explores the tracking modal parameters for bridge anomaly warning.The regression analysis model between modal parameters and environmental factors is constructed,and the control chart is used to control the modal parameter deviation caused by identification methods,which are used to reduce the false alarming rate caused by the modal parameter variation of bridge in health state.The early warning results based on the tracking modal parameters in different vibration forms are analyzed,which provides a reference for the early warning of real bridge anomalies.