Pattern Analysis And Model Prediction Of Bridge Health Monitoring Information

R&D of bridge health monitoring become an active domain of academics and engineering all over the world. Monitoring data are mass, so analysis of monitoring data play a important role in the research work, which must be accurate and immediate. Assessment and prediction to safety of bridge are established on the base of analysis of monitoring data. Pattern analysis and model prediction of bridge health monitoring information are discussed in this dissertation, which are based on the real-time health monitoring system of Qianjiang 4th Bridge.Firstly,The real-time health monitoring system of Qianjiang 4th Bridge is introduced, according to the analysis of one-year-long monitoring data of the system, this dissertation apply power spectrum density analysis for Structural Dynamic characteristics to gain its formant frequency, Which is compared to the results of FEA, as a rusult ,the good agreement is shown.Secondly,this dissertation apply statistical pattern analysis for strain data of arch feet, tension data of cable,flexibility data of straining beam and establish their temperature pattern,wind velocity pattern and time pattern,year pattern is established for strain data of arch feet.From the pattern ,we can see the changeable regularity of the monitoring data. which is different form change caused by damage and is favorable for assessment of bridge safety.Finally,this dissertation has established AR medol,ARMA model and ARIMA model,GM(1,1) model for partial monitoring data based on time series analysis and grey theory, predictions have been given by using those models. The results are compared, GM(1,1) model need little information and its short-term prediction is better than time series model, but,the longer term prediction ,the worse results for all models.