The Structural Damage Identification Theory And Application Based On Root Mean Square Difference Method

Bridge damage identification and health monitoring is recently a hot point in the world. The bridge is an important part in the transportation and economic development for the country, and the bridge damage detection can ensure the safety of the bridge. The technique of damage detection based on vibration is to evaluate the bridge's condition using vibration signals. This technique is expected to find little damage and identify the position of the damage before dangerous events happen. And also, it can help people learn about the condition of the bridge, save repairing expense, reduce the probability of lost, and provide the proof of repairing. This paper includes three parts: structural damage identification theory, damage identification experiment of simple-supported bridge using root mean square difference method based on artificial neural network, the analysis of a bridge model damage identification using this method.This paper presents a damage evaluating method-root mean square difference technique to evaluate the health condition of the structure. For a good structure, impulse response signals are obtained using natural excitation technique in environmental vibration condition. The impulse response signals are used to train neural network as the in and out data. The trained network saves the information of the structure vibration. When some damage happens or environmental condition changes, the recent structure signal is imported to the former network, then the out-data will differ from the in-data because the network is for the former structure. The root mean square of the difference between in and out data is defined damage index, root mean square difference. If the damage position and grade change, this index will change accordingly.In the reinforced concrete simple-supported beam loading experiment, vibration experiment is made and the neural network method could identify the position and the grade of the damage clearly. The identification results are similar to the results of the numerical model.Time history analysis is applied in the model of Yinpenling bridge in Changsha. Impulse response series are obtained from the Natural Excitation Technique. The neural network is established using the series and damage is identified in different damage degree condition.