Structural Damage Identification Under Changing Environment Based On Cointegration Method

The safe operation of civil engineering structures is closely related to the safety of people’s lives and property.In order to find structure damage as early as possible and ensure the safety and reliability of structures,the Structural Health Monitoring（SHM）system emerges at the right moment.In the field of SHM,how to effectively analyze the monitoring data of the health monitoring system to realize the estimation of the damage state of the bridge structure is a core task of SHM.In recent years,many damage identification methods have been proposed by related researchers.However,these methods still have many shortcomings when applied to actual engineering structures.The change of environmental factors and operational conditions is a very important factor that restricts their application,and it is also a big problem.Therefore,considering the impact of the change of environmental factors and operational conditions,this thesis conducts research on strutual damage identification,and carries out the following research work:（1）This thesis introduces the influence of environmental factors on the dynamic characteristic of structures,summarizes the common structural damage identification methods considering the influence of environmental factors,and expounds the benefits and drawbacks of these methods.（2）A nonlinear clustering cointegration method is proposed to overcome the defect that the traditional cointegration method is only applicable to the linear correlation between monitoring variables.In this method,Principal Component Analysis（PCA）and Gaussian Mixture Model（GMM）were used to transform the bilinear relationship into two linear relationships,and then the cointegration analysis of the two linear relations is carried out respectively.The residual sequence obtained from the cointegration is used as the damage factor without the influence of environmental factors,and the damage status of the structure is judged by the X-bar diagram.The numerical model of spring mass systems of a 4-DOF,a 16-DOF and the measured data of Z24 bridge are used to verify the effectiveness of the proposed method in dealing with the nonlinear monitoring variables.（3）A two-step PCA method for environmental factors sepraration is put forward,that is,before traditional PCA is used to identify the damage,clustering PCA was carried out on the frequency of the structure to transform the bilinear relation into two linear relationship,and then PCA is performed on the linear variables respectively.Finally,the Euclidean distance is used as the damage index that are considered to have separatd the influence of environmental factors.The validity of the proposed method is verified by a 4-DOF and a 16-DOF numerical models and the data of Z24 bridge.（4）A cointegration pair method is proposed to distinguish the data anomalies caused by sensor faults and structural damage.Taking a prestressed concrete beam numerical model as an example,the method put foward can distinguish and identify structural damage and sensor fault while separating the impact of environmental factors on the data.The results show that the method can effectively identify the moment of occurrence of the structural damage and sensor fault,and determine the sensor fault location by analyzing the deflection data of the prestressed concrete beam.In addition,it can be found that this method can effectively judge the sensor fault and its location by analyzing the actual data of a Yangtze River bridge.