Jie Liu, Han Cheng ,Qingkuan Liu,Hailong Wang and Jianqing Bu6



To obtain an effective data mining method for cable-stayed bridge damage diagnosis, the algorithm of the cable-stayed bridge damage diagnosis model based on data mining was studied, and a data mining method is proposed.

This method is oriented to the damage diagnosis of cable-stayed bridges. After algorithm comparison, the support vector machine (SVM) and limit gradient-boosting (XGBoost) algorithms, with advantages in damage location and quantification, are combined and optimized to obtain the damage diagnosis model for cable-stayed bridges.

First, a refined benchmark finite element model is established by Abaqus, and postprocessing data such as vibration frequency and modal curvature are used as a data mining dataset. Second, feature se-lection is conducted, and the damage-sensitive modal curvature change rate index is selected as the feature of data mining.

Next, the SVM and XGBoost algorithms are optimized by grid and random search, and the optimized SVM and XGBoost algorithms are used to locate and quantify the damage. Finally, the damage diagnosis model for cable-stayed bridges is obtained. Taking a cable-stayed bridge as an example, the proposed method is applied and analyzed, and the results show the effectiveness of the proposed method.



damage diagnosis of cable-stayed bridges; damage diagnosis model; data mining; python; support vector machine; XGBoost