桥梁结构监测主要集中在对桥梁结构损伤位置和损伤程度的研究,然而,这是以桥梁结构健康监测系统数据有效为前提的.在实际的环境里,由监测系统自身故障引起的异常往往会对监测数据有一定影响,使得监测系统产生损伤误报,增加了虚警率;同时,由某些外部荷栽引起的突发事件,可能会对结构有严重破坏,不利于桥梁的安全维护和管理.为了保证桥梁的安全,提高桥梁结构监测的有效性,有必要对特殊事件进行异常诊断.该文将一类识别方法应用到桥梁数据诊断中,即通过核主成分分析和超球面一类支持向量机方法将一般监测数据和特殊事件数据有效区分,并利用江阴大桥的加速度传感器数据验证了该方法在船撞、台风、传感器装机噪声和传感器跳变信号下的有效性。
The research on the bridge structural monitoring focuses mainly on the identification of the structure damage position and degree. However, these researches are based on the data of the structural health monitoring system. In the actual environment, the abnormal data caused by the failure monitoring system can often make the false prognosis, increasing the false alarm rate. Meanwhile, the bridge may have serious structural damage from the unexpected events caused by some external loads. They are not conducive to the bridge safety maintenance and management. In order to ensure the bridge safety and improve the effectiveness of the bridge structure monitoring, it is necessary to diagnose these special events. In the paper, kernel principal component analysis (KPCA)and hyperspherical support vector machine method are employed to separate the general monitoring data from the event data. The acceleration sensor data in Jiangyin Bridge is used to validate the effectiveness of the method under the ship collision, typhoons, sensor installed noise, and sensor step signals.