Time-domain Damage Identification Of Bridge Based On The SVM Kinetic Energy Indicators

This thesis through summarizing and analyzing the damage identification bridgestructure on time-domain method found that the research about the influence ofbridge damage under vehicle loads during the bridge operation is less. On thataccount, this thesis will use the velocity information which is obtained from thedynamic response of bridge structure in time domain for constructing a kinetic energythan as the damage index, and then used the ratio of change in kinetic energy than foreach element after occurrence of the damage for damage localization andquantification in bridge structures. Applying this method to the finite elementnumerical simulation experiments, the main research work includes:（1）Through read a large number of domestic and foreign literatures, this thesissummarized and analyzed the background and significance of bridge structuraldamage identification, its development history and research status quo. It introducedmain optimization identification method and pattern recognition method for theapplication in time-domain damage identification of bridges. And analyzed theexisting problems on application of these methods in the bridge damageidentification.（2）It is expounded that the time-domain damage identification method basedon SVM kinetic energy index, included the basic theory of support vector machine(SVM), the basic concepts and advantages, the maximum margin separatinghyper-plane, kernel function, the classification of the margin and the SVM algorithm,the kinetic energy than the damage identification theory. Making use of the dynamicequation which belongs to simply supported beam under the moving load and derivedthe dynamic response of the beam with forced vibration on bridge-vehicle couplingsystems in a certain time, the use of the dynamic response data build the kineticenergy than indicators for damage localization and quantification in bridge structures.（3）Using the above method through the numerical simulation experiment.Firstly we introduced experimental methods and procedures, given concrete damagemodel and car lane load implementation approach. The establishment of finite element model of simply supported beam is discussed the implementation processwhich the time-domain damage identification methods based on the SVM kineticenergy index, and then for studying the applicability of this method, expanded andapplied the method on a certain proportion shrinkage finite element model which isthree span prestressed concrete continuous girder, used the same method to identifythe damage localization and quantification in the bridge.（4）Time-domain damage identification of Xu-Ge bridge. Through the largespan Cable-stayed bridge—Xu-Ge bridge as the research object, a practical Xu-Gebridge model with the optimal placement of the sensor data as the study foundation,use the kinetic energy index SVM method was verified the feasibility of damageidentification, achieved good results. For the damage identification of the main toweris given some probes, at the same time, due to the research of this method is theorystudy currently, we needs further research to make it really applied to the bridgestructural health monitoring.