Research On Monitoring Bridge Condition And Train Operation Parameters Based On Distributed Macro-strain

Bridges are important infrastructure for high-speed railway bridges,with the characteristics of spanning complex terrain and saving floor space.In the context of the increasing mileage of high-speed railway construction,bridges with different structural forms are more frequently used in the line,with simple supported beam bridges with different standard spans being the most common.With the gradual transformation of high-speed railway lines from large-scale construction to the stage of operation and maintenance,it is very necessary to monitor bridge health and train operation parameters.This paper focuses on the health monitoring system for railway simply supported beam bridges based on long gauge distance FBG strain sensors.Firstly,a finite element model of train track bridge coupled vibration is established using the displacement contact method,and the track plate coupling and bridge beam macro-strain solution methods are verified;Secondly,comparing and studying the dynamic and static separation methods of macro-strain time history signals obtained from long gauge distance FBG sensors under train track bridge coupling vibration.For simulation signals,wavelet decomposition,EMD method,and improved EMD method are respectively adopted for signal processing,and the advantages and disadvantages of each method are compared.The fastest and accurate signal processing method is selected and used for the actual signal processing of the bridge health monitoring system;Further,a method for monitoring the longitudinal bending stiffness of bridges based on macro-strain influence lines is studied.Under the action of random train loads,the longitudinal bending stiffness monitoring method is verified under three conditions: single,multiple,and full bridge range stiffness degradation,and the method is applied to the stiffness monitoring of actual bridges;Finally,a train operation parameter identification method based on macro-strain influence line and BP neural network is studied,and the impact of sample set division and identification parameter types on the network recognition effect is explored.The main contents and conclusions are as follows:(1)The method for solving macro-strain of bridges under train track bridge coupled vibration is studied.A finite element model for solving train track bridge coupling vibration is established using the displacement contact method using APDL programming under the ANSYS environment,and the vehicle track bridge coupling method is studied.According to the dynamic principles of the three components of train,track,and bridge,the method of establishing contact pairs between wheels and rails is used to establish a connection between the train and track bridge.The inverse Fourier transform method is used to obtain track irregularity samples and import them into the track nodes.Finally,the verification of the coupling between the track and track plate and the verification of the macro-strain solution method for bridge beams are carried out,which proves the reliability of the solution model established in this paper,it lays the foundation for the following research content.(2)The macro-strain signal processing method under train track bridge coupled vibration is studied.An improved algorithm based on the EMD algorithm is proposed.In the MATLAB software simulation platform,wavelet decomposition,EMD method,and improved EMD method are respectively used for dynamic and static separation of macro-strain signals.By comparing the signal to noise ratio(SNR),root mean square error(RMSE),and Pearson correlation coefficient(R)as indicators,it is found that the wavelet decomposition of the separated quasi static signal quality is the best,followed by the improved EMD method and EMD method,processing time and stability are the most advantages of the improved EMD method,and wavelet decomposition relies excessively on manual selection.(3)The method of bridge stiffness evaluation and early warning based on macro-strain under train track bridge coupled vibration is studied.Firstly,the health monitoring system for a simply supported beam bridge across a high-speed railway installed by the research team is introduced;Then,the two characteristic parameters of train carriage mass and vehicle speed are statistically analyzed and a random train load model is constructed;Then,a method for evaluating the longitudinal bending stiffness of bridges based on the macro-strain influence line was studied,and a health confidence interval was established to match the bridge with random train loads.The stiffness degradation was located and quantitatively analyzed through the area increment of the macro-strain influence line of each sensor.The feasibility of this method was verified by numerical simulation examples,with significant results,based on the bridge health monitoring data,it was identified that there was a sudden change in the box girder section at the location covered by the sensor gauge distance,resulting in a sudden decrease in the longitudinal bending stiffness of the section.(4)A method for identifying train operation parameters based on distributed macro-strain monitoring is studied.From the point of view that the macro-strain time history curve contains information about train operation parameters,combined with the area of the macro-strain influence line,a BP neural network for identifying train operation parameters is constructed and trained.The research results show that when identifying the car mass,the relative error of all samples is within 2%,and the qualification rate is 100%;When identifying the wheel/rail force curve,for a single identification sample,the wheel/rail force numerical error at each time point is within 4%.For the entire test set,the relative error of the network is 3.843%,and the pass rate is 96%.