Prediction method of bridge static load test results based on Kriging model

摘要

To solve the problems of expensive bridge load test cost, traffic congestion influence, and damage to bridges by load test, the dynamic experimental results based on the inexpensive, safety inspection data and maintenance process of existing bridges are presented in this paper. The Kriging model is used for the intelligent analysis and prediction of the actual stiffnesses of the existing bridges as well as for the high-accuracy prediction of their static load experimental results. In order to achieve the above objectives, a sensitivity coefficient is selected based on the sensitivity analysis of the whole bridge, and the Kriging model is established and optimized to forecast and modify sensitive parameters for the high-precision correction of the model. Relative to other machine learning algorithm models, the Kriging model has higher parameter sensitivity and reliability. To verify the correctness and feasibility of the above mentioned methods, a continuous rigid frame bridge is selected as an engineering test object, and ANSYS, a finite element software, is used for modeling and analysis. The research results show that the finite element model modification method based on the Kriging process can employ inexpensive and convenient bridge dynamic load tests to modify the actual parameters of the finite element model in the Kriging process of bridges; consequently, the test results of static load experiments can be more accurately predicted. The correction results obtained by the Kriging model are in good agreement with test results and exhibit high precision and reliability; moreover, the method is less costly and good safety, and has minimal influence on traffic. Moreover, in view of the potential for conducting a large number of bridge mechanical performance evaluations on all levels and the effective simulation and performance prediction of existing bridge project maintenance decisions, the proposed method affords a new train of thought.