Research On The Existing Bridge Structure's Health Diagnosis And Remaining Life Prediction

Research on the in-service bridge structure's health diagnosis and its remaining life prediction is a necessary part in bridge conservation, maintenance and technical reconstruction today, and it has become a hot research at home and abroad. With the increment of old bridges in our country, the problem attracts more and more attention. Combining with the character of bridge, this paper gives a thorough study on some technical problems of bridge structure's health diagnosis and its remaining life prediction, which includes:(1) The basic theory, modeling process and solution feature of grey forecasting model, grey correlation degree and particle swarm optimization algorithm are concluded, each of the models'mainly suitable condition is analyzed, as well as lay a foundation for their application in the bridge structural damage assessment and life prediction.(2) A fixed-end beam's finite element model under dead loads with different assumed damage conditions are computed, the parameter choice of bridge detection and sensor placement based on damage sensitivity are analyzed. Compared with the computed results, it shows that the strain parameter is more sensitive than the displacement parameter under the assumed damage conditions. At last, combining the finite element model's analyzing results which based on damage sensitivity and the results of sensor placement based on experience, the model bridge's sensor is primarily distributed, it gives some reference to the bridge detecting.(3) Based on the curvature modal grey correlation degree and the GM(1,1) model based particle swarm optimization, a girder bridge damage detection technology research is conducted. Taking a concrete girder bridge's finite element modal simulation as an example, the method of curvature modal grey correlation degree is used to identify the damage location under different degree of injury and damage working condition. The analysis shows that the method have a good recognition accuracy and sensitivity of girder bridge's single injury and multiple injuries under dead load, meanwhile, it uses the low-modal displacement values which can also be easily obtained in practical engineering. With frequency as a bridge structure's damage identification parameter, the GM(1,1) model based on particle swarm optimization is used to identify its damage. Compared with the original GM(1,1) model, the new model not only has high recognition accuracy, but also has obvious advantages in the non-flat data treatment, it is more suitable for bridge structure's damage identification which has complex structure and working conditions.(4) The GM(1,1) model based on particle swarm optimization is introduced into remaining life prediction for in-service bridge, the analysis shows that the method has high prediction accuracy, and it can be better used in the treatment of volatile data than the original GM(1,1) model, so the prediction would be more feasible and rational when it is applied to the real bridge.