中文摘要：

监测信息的数据失真识别是桥梁结构长期健康监测系统的核心问题之一。针对桥梁结构长期健康监测系统监测信息的海量数据中由于仪器设备失效或环境干扰而导致的数据失真问题,结合数据变化率和聚类变化等数学统计方法,提出基于数据变化率的识别算法。该方法主要是对比数据之间的结构特性,首先建立相应结构参数的结构特性数据库,然后训练出该类型参数的安全系数以及变化率阈值,并通过数据库的不断更新和扩大来提高安全系数和阈值的精确性,最后通过验算实时采集数据的变化率是否超过对应的样本阈值来判别该数据是否失真。通过对佛山平胜大桥长期健康监系统采集的主塔偏位和环境温度数据进行失真识别验证,表明了该算法在实际工程中运行的有效性和可靠性。

英文摘要：

Monitoring data distortion recognition is one of the key issues of the long-term health monitoring system for bridge structure. According to the problem of data distortion in the massive data of monitoring information collected by the long-term health monitoring system due to equipment failure or environmental disturbance, combined with the data changing rate, clustering change and other mathematical methods, we put forward the data distortion recognition algorithm based on data changing rate. This method is mainly to contrast the data structure characteristics： First, to establish the structural properties database corresponding to the structure. Then, to train the safety factor and changing rate threshold of such kind of parameters, and to improve the accuracy of safety factor and threshold by constantly updating and expanding the database. Finally, to judge the real-time data＇s distortion by checking whether its changing rate exceeds the corresponding sample threshold. By verifying distortion recognition of the data of pylon displacement and ambient temperature of Foshan Pingsheng bridge which collected by the long-term health monitoring system, it shows that the algorithm is effective and reliable in the actual project.