Research On The Damage Identification Of Girder Bridge Structures Based On Auto-Regressive Moving Average Model

The damage identification of bridge structure is an active research field in the current bridge engineering, also has a broad background, theory and technology, related to this field, has being continued to develop and mature. The research method based on Auto-Regressive Moving Average (ARMA) model is often used in state predicts or system identification in the past and it is still in the development and perfect stage to use this method in the huge bridge structure. According to the feature of bridge, a further research on the application of this method in the damage identification of bridge structure will be discussed. The main tasks of this thesis are as follows:1. referenced to the concepts of structural damage identification, the current research situation and development trend of damage identification is described concisely. Basic concepts such as system identification etc and MATLAB language that are involved in the method of damage identification based on Auto-Regressive Moving Average model are introduced briefly.2. the basic principle of time series analysis is introduced concisely, then the time-domain characteristics of ARMA model, including the Green function, revelation function, autocorrelation function and partial autocorrelation, is introduced briefly. All of those are the theory basement of the establishment of ARMA model.3. the modeling process of ARMA model is introduced in details, including data collection and standardization, estimation of model parameters, determination of the order of model, and so on. The estimation of model parameters and determination of the order of model is the key to build the ARMA model correctly. In order to estimate parameters of the model, long-autoregressive method and ordered estimation method are introduced. In order to determinate the order of the model, residual square criteria, F-criteria, and AIC criteria are introduced.4. the intrinsic relation between time series and system model is introduced, then, the equivalent relation between ARMA model and vibration system is introduced briefly.5. there are numerical verifications about damage identification methods mentioned in this thesis.Taking a simply supported beam bridge and a continuous beam bridge as the example to build structural finite element model with ANSYS procedures, according to the dynamic analysis to get the acceleration data of each node under the white noise excitation. Then, build ARMA model according to MATLAB procedures, considering the different level noise impaction, thus verify the feasibility, effectiveness and applicability of this damage identification method based on ARMA model.Finally, the main conclusions of this work are figured out and some subjects for further research are also discussed.