Bridge Damage Detection And Optimal Sensor Deployment Based On The Analysis Of Deflection Influence Line

Bridge structures are exposed to the effects of insufficient design,construction defects,heavy traffic,overloaded,earthquake,shock and corrosion,hence,structural damage will be accumulated during the whole life cycle.Detecting and repairing damage in time are of great significance for extending the service life of bridge structures as well as guaranteeing the safety of people’s life and property.Vibration modal and its derivatives-based indicators are utilized most frequently in bridge damage detection,which mainly employ single point excitation,multi-point output pattern to provide damage detection with more useful information.Another kind of approach to settle the insufficient information problem is multi-point excitation,single point output pattern,namely,reflecting the mechanical performance as well as local anomaly of structures by analyzing the response of bridge under unit moving load.In term of the bridge structure bearing moving vehicle load,the second pattern is more economic and convenient for practical application.Hence,bridge damage detection and optimal sensor deployment based on the analysis of deflection influence line had been systematically studied in this thesis from three aspects,namely,1)influence line extraction based on structural response under moving vehicle load;2)damage localization and damage extent quantification based on deflection influence line;3)optimal sensor deployment based on damage sensitivity analysis of influence line.This thesis proposed a cubic B spline curve and Tikhonov regularization-based influence line extraction method,which using vehicle parameter including wheel base and axle load and the dynamic vertical displacement response of key section in time domain as input,and then,cubic B spline curve was utilized as constraint condition to improve the smoothness of influence line.A four-span cross section continues beam simulation model and laboratory experiment were conducted to verify the feasibility of proposed method,results show that the influence line extracted by the proposed method not only agree well with the baseline,but also meet the smoothness requirement of influence line.Within the damage localization problem,four kinds of deflection influence linebased indicators had been summarized in this thesis systematically.A damage localization method based on the analysis of influence line by using wavelet transform.Hereafter,the concept of influence line extends to influence surface and 2-D wavelet transform had been adopted to realize the damage localization of elastic plate.Results of numerical example show the effectiveness of proposed method in small damage localization without depend on influence line baseline in health state.Aiming at the problem of damage extent quantification,a hyper-static matrix equation about stiffness change matrix,damage coefficient matrix and deflection change vector had been deducted.Then,damage extent of each element could be determined by limiting the bound of damage coefficient and solving the generalized inverse matrix equation.Both numerical example and laboratory experiment had been conducted to verify the effectiveness and feasibility of proposed method.The effects of multi-damage,sensor location,the number of sensors and noise level had been analyzed thereafter.An influence line-based damage detection oriented optimal sensor deployment method had been proposed to improve the limited damage detection ability bring by single sensor.The method constructed Fisher information matrix from damage sensitivity matrix to calculate the contribution of each node to damage detection.Distance correction coefficient had been introduced to Fisher information matrix to solve the information redundancy problem bring by traditional effective independence method.Results show by a three-span continue beam model indicate that sensor deployment result given by the proposed are more decentralized than traditional one,and all the sensor had been set in node which make significant contribution to damage detection,additionally,the overall damage detection relative error is significantly low than that of traditional method.