Research On Fatigue Inspection And Maintenance Optimization Of Steel Bridges Based On The Dynamic Bayesian Network

With the rapid development of bridge construction,China has a large number of bridges.As a main form of bridges,steel bridge is prone to fatigue damage under the repeated actions of vehicle and wind loads.In order to achieve sustainable development,not only to ensure the safe operation of the bridge structure in the whole service life,but also to make the maintenance cost lowest,the inspection and maintenance optimization of the steel bridge should be studied.In this paper,using the inference function of the dynamic Bayesian network and combing the relevant knowledge of fracture mechanics,the fatigue crack growth model of steel bridge based on the dynamic Bayesian network is established and completed by applying the Ge NIe software.The model’s inference and prediction functions are applied to predict the crack length of fatigue details,to consider the influence of maintenance costs,and finally to make a decision on the inspection and maintenance behavior.The main research contents and results are as follows:(1)The construction process of the dynamic Bayesian network is studied.Combined with the relevant knowledge of fracture mechanics,the fatigue crack growth model of steel bridge based on the dynamic Bayesian network is established by using Ge NIe software.(2)According to the function of the dynamic Bayesian network,the annual overrun probability of steel bridge in service life is predicted,and the state of fatigue details is evaluated based on the overrun probability.(3)Taking the orthotropic steel bridge deck as an example,three common fatigue details are analyzed,and the overrun probability of fatigue details during service period is studied;Taking the weld of longitudinal ribs and diaphragms as an example,the application of dynamic Bayesian network in crack propagation is studied through the forward inference,setting evidence and reverse inference functions of dynamic Bayesian network,and the parameters of the node variables are analyzed.(4)The inspection probabilities of three common nondestructive inspection technologies are studied,and a new method for determining the nondestructive inspection technologies is proposed in combination with the fatigue crack growth model of steel bridge based on the dynamic Bayesian network.(5)Based on the probability method,the maintenance cost is integrated into the decision-making network,and the benefits of non-destructive inspection technology under different inspection results are analyzed.(6)A time-varying fatigue detail crack maintenance threshold is proposed according to the maintenance cost and failure cost.When it is determined that maintenance measures need to be taken,a method for determining the optimal maintenance time is proposed.