ABSTRACT

Practical application techniques are really required in bridge structural health monitoring (SHM), while exploration of structural damage detection has made great achievements in recent years. Based on the measured structural dynamic responses from monitoring system in a serving bridge, this paper utilizes statistical indicators from huge amount of monitoring vibration data and a single-class Support Vector Machine (SVM) to build a special pattern recognition algorithm and achieve the goal of operating status pre-warning for the bridge. In order to obtain an effective input vector for the SVM algorithm, acceleration response data from a arrange of measuring points on the main girder and arch ribs is statistically analyzed. Consequently, three statistical indicators, namely mean, variance, and 6 quantile points values of daily acceleration record, are selected to establish an eight-dimensional vector as the input information of the single-class SVM for structural damage identification. The proposed method is certified to be feasible with the application example in an operating steel arch bridge.