Study On Bridge Structural Parameters Identification And Damage Detection Techniques Based On Bridge-vehicle Coupled Vibration Analysis

For the safety, applicability and durability of bridge structures, the defects or damages of bridges must be detected in time, In the traditional damage identification methods there are many drawbacks, where the system excitations are difficult to implement and measure because they influence the normal bridge operation, too many measuring points are required, measuring information and data are incomplete to reflect the dynamic performance of the bridge in working state, and so on. According to these considerations, this thesis studies a bridge parameter identification and damage detection method using operating train loads as excitation sources. The work focuses on the following aspects of research:1. The bridge structural damage identification techniques based on bridge-vehicle coupled vibration analysis are summarized and compared with the traditional methods. Their advantages and disadvantages are investigated, in which the working principles of five methods are introduced in detail. The procedures and the strategies in practical application of these methods are summarized.2. A dynamic analysis model of a simply-supported beam under a vehicle model of wheel-spring-damper-sprung mass is established, and the motion equations are derived, in which the influences of deck irregularity and vehicle speed-varying motion are considered. Using the vertical accelerations of vehicle and its spectra, the fundamental frequency of the bridge is identified. The computer code LBV for bridge fundamental frequency identification is developed. By some numerical examples, the influences of parameters on the identified results are studied, such as bridge damping, vehicle spring stiffness and damping, mass ratio of vehicle to bridge, vehicle moving speed and acceleration, and bridge surface roughness.3. The coupled dynamic analysis model for the bridge-train system is established, in which the train model consists of several vehicles with each having 10 DOFs, and the bridge model with beam elements. The computer code BVD for the analysis of dynamic interaction between bridge and train is developed, and the dynamic characteristics of the bridge before and after damage are studied and compared.4. A transient data-based sensitivity method for bridge damage identification is proposed, which uses the measured dynamic responses of bridge excited by moving trains or dynamic responses of vehicle, and the sensitivity analysis of dynamic response to the calculated bridge damage index. In this method, by modifying the damage indices in each iteration step to update the bridge finite model, the damage indices can be finally identified with the condition that the measured response accords with the calculated one. In addition, the damage identification method using only partial vehicle parameters is proposed. The principle of Tikhonov regularization method is discussed in detail, and the application of the method in bridge damage identification is studied. The computer code BVDS for damage index identification based on this method is developed, and the feasibility of the method is validated by numerical examples. The influences of dynamic response types, measurement point locations, track irregularity, measurement noise and train speed on the identification errors are studied.5. The signal threshold value de-noising theory, method, procedures and strategies based on wavelet analysis are discussed. The influences of wavelet function selection, wavelet threshold evaluation regulation and threshold function on de-noising results of bridge dynamic responses excited by trains are studied. The de-noising effect is verified by using the de-noising signals for damage identification.