提出了考虑温度变化影响的悬索桥结构损伤预警方法.首先,采用神经网络技术建立桥梁实测模态频率与温度的相关性模型,用以消除温度变化对模态频率的影响.然后,将不同温度下的实测模态频率进行"温度归一化",在此基础上利用神经网络新奇检测技术建立自联想神经网络进一步识别模态频率的异常变化.通过润扬大桥悬索桥236d的实测数据分析验证了该方法的可行性.分析结果表明,不同季节下模态频率的相对变化平均约为2.0%,采用所提方法可以识别出悬索桥模态频率0.1%的异常变化,适用于悬索桥结构的在线整体状态监测.
This paper aims at successive structural damage detection of long-span bridges under changing temperature conditions.First,the frequency-temperature correlation models of bridges are formulated by means of artificial neural network techniques to eliminate the temperature effects on the measured modal frequencies.Then,the measured modal frequencies under various temperatures are normalized to a reference temperature,based on which the auto-associative network is trained to monitor signal damage occurrences by means of neural-network-based novelty detection techniques.The effectiveness of the proposed approach is examined in the Runyang Suspension Bridge using 236-day health monitoring data.The results reveal that the seasonal change of environmental temperature accounts for variations in the measured modal frequencies with averaged variances of 2.0%.And the approach exhibits good capability for detecting the damage-induced 0.1% variance of modal frequencies and it is suitable for online condition monitoring of suspension bridges.