Study On Damage Identification And Assessment Of Small And Mid-Span Bridges Based On Long-Gauge FBG

Small and medium span bridges are widely employed in highway system.As traffic load continues to increase,their properties would deteriorate more or less over time.In order to detect and avoid potential disaster in time,a structural health monitoring(SHM)system can be installed to timely monitor and assess a bridge's property.However,for current SHM systems,much attention has been attached to the long span bridges instead of small and medium span bridges.Compared with conventional sensors,the long-gauge FBG sensors developed by our group can monitor structural response within a large region only by several sensors.With advantages on installation convenience,these sensors will have a bright prospect in future application.Based on this type of sensor,a sensing system has been established aiming to realize multiple damage identification and assessment functions required in SHM system and to meet the specific characteristics of small and medium span bridges.The main contents of this study are including:A bridge weigh-in-motion system based on difference between long-gauge strain responses is developed.On the basis of the conception of strain influence line,a formula of long-gauge strain influence line was deduced.According to the relationships between variance value among three consecutive long-gauge strain signals and vehicle's velocity,wheelbase and axial weight,a strain index?_D was proposed to construct a bridge weigh-in-motion method free of the limitation of structural boundary condition.Then a series of numerical simulations and actual bridge test were conducted to testify this method's feasibility.Then,a damage identification method on the basis of long-gauge strain‘s multi-cross reference for small and medium span bridge was proposed.Under a situation with single vehicle passing,a function relationship between random three long-gauge strain responses and structural damage was deduced to establish this damage identification method.This method utilizes multiple combinations of entire signals obtained by the sensors to carry out multi-cross reference.The damage location and extension can be accurately identified and effectively eliminate the false alarm conditions caused by sensor malfunction and external environmental interference.Then based on vehicle-bridge coupling simulation theory,a two dimensional vehicle-bridge coupling simulation program was programmed to validate this method's feasibility with a series of numerical simulation scenarios being operated.The performance under actual environment was also tested on an actual bridge.After that,another damage identification method for small and medium span bridge under long-gauge strain time histories caused by stochastic traffic flow was proposed.According to the relationship between the integral of long-gauge strain influence line and structural stiffness distribution,the integral of long-gauge strain time history generated by stochastic traffic flow was derived.It is found that the relationship between this integral and stiffness distribution is approximately equal to the one under single vehicle scenario.Under this finding,this damage identification method was developed which is functional for actual stochastic traffic condition instead of being limited in single vehicle passing conditions.And a scaled bridge model was utilized to build an indoor vehicle-bridge coupling experimental platform to conduct a series of indoor experiments.After that,to testify the performance under more actual traffic condition,a three dimensional vehicle-bridge coupling simulation program was compiled to generated stochastic traffic flow based on actual monitored data.Based on former proposed two damage identification methods,a two stage model updating framework based on long-gauge strain time history was established to updating the important parameters in a finite element model including material property and boundary conditions.The updating method utilizing Bayes'theorem can obtain each parameter's posterior distribution by combining their prior distribution obtained through damage identification method with Markov chain Monte Carlo method.In model updating,several trained radial basis function networks were used as surrogate model to replace the finite element model to reduce computing time during model updating process for timely updating and assessment of the bridge.Later,a series of numerical simulations and indoor experiments were carried out to validate this method.At last,in this series of studies,a bridge reliability assessment method based on long-gauge strain and probability density evolution method(PDEM)was developed combining all mentioned study results above.The main idea is by combining aforementioned updated finite element model as foundation with stochastic traffic flow model and PDEM.Through calculating and analysing structural random responses,the current reliability condition of the bridge can be assessed instead of using Monte Carlo method to calculate reliability resulting in low efficiency.A typical highway small span bridge with an actual monitored stochastic vehicle load model was utilized to build a numerical case to operate fatigue reliability assessment considering model's material and boundary uncertainty under different traffic conditions to illustrate this method's feasibility.