A Study On Key Issues Of Bearing Capacity Evaluation For Bridge Structures Based On Health Monitoring Systems

The bearing capacity evaluation of the existing bridges is deemed as one of the ultimate goal of the structural health monitoring(SHM),rising a great research interest in the SHM community.To realise the evaluation,the probabilistic-based damage identification and the robust estimation of the traffic load effect,which stand as two premises,should be well addressed beforehand.Meanwhile,the main objective is to develop a reliability-based bearing capacity evaluation method for existing bridges which can ideally connect/link with the structural bearing capacity calculation under the proposed(design)stage.Several critical issues contained in the evaluation are studied in this dissertation and the main works is listed as follows:(1)Literature review on the recent development of structure health monitoring and the bearing capacity evaluation of existing bridge structures.The concept,system architecture and application status of bridge structure health monitoring are introduced.Then the progress of the key technologies for the evaluation are reviewed,including the damage identification tools,the traffic load modelling issue and the evaluation method.A novel structure health monitoring system is proposed with the application to the Xinguang Bridge.(2)Research on time-series-based probabilistic structural damage identification.The conventional time series analysis method can only verify the occurrence of the damage but often fails to probabilistically characterise the damage after it is detected.In this dissertation,three damage indexes,i.e.the residual index,the first three steps and the Euclidean distance,are proposed and the distributions of three indexes are studied.The probabilistic damage calculation algorithm along with the damage threshold are established based on the damage indexes,which offer a probabilistic measure of the damage occurrence.A new evaluation method for the sensitivity and robustness of the damage indexes in terms of the aspect of damage probability are presented,following the numerical studies which are used to evaluate the performance of the three proposed damage indexes.Finally,with the use of the real-world monitoring data acquired from both experimental model and the Xinguang Bridge,damage identification is conducted to verify the effectiveness of the proposed method.(3)Research on robust Bayesian estimation of traffic load effect.The traditional statistical analysis of the traffic load effect treats the model as fixed which ignores the uncertainty contained in the model estimation and model selection.By pointing out the uncertainty contained in the measurement of the traffic load effect,this dissertation proposes a Bayesian estimation method on traffic load effect section distribution with which both parameter uncertainty and model uncertainty can be explicitly quantified.Through the model weights and the posterior parameter distributions,a more robust estimation for the traffic load effect can be obtained.The proposed method is further applied to model the traffic load effect section distribution of the Xinguang Bridge,and the maximum load effect in the target operation period is extrapolated and updated.(4)Research on the reliability-based evaluation method for bearing capacity of the existing bridges.In order to connect/link with the bearing capacity calculation of the newly designed bridge,a yearly failure probability equivalence method is utilised to update the target reliability index for future operation of the existing bridge.Based on the damage identification results and the manual inspection information,the statistic parameters for the deteriorated resistance are determined,meanwhile,the traffic load effect distribution is updated by the robust Bayesian estimation.With the use of the updated time-varying values of the reliability index,the resistance and the traffic load,a resistance-based evaluation method in the context of reliability theory is established,given an explicit utilization rate of resistance which suits for engineering practice.The proposed method is further used to assess the bearing capacity of the Xinguang Bridge,verifying its feasibility and effectiveness.The research outcomes indicate that 1)the time-series-based probabilistic damage identification method can provide probability measure of the damage occurrence;2)the Bayesian-based method can offer a robust estimation on the traffic load effect model;3)the reliability-based evaluation method in conjunction with the updated resistance and load information can give a more reliable evaluation on the bearing capacity of the existing bridges.The successful application of the new methods in the real-world Xinguang Bridge verifies the effectiveness and feasibility in the large-scale bridges.