中文摘要：

基于环境条件变化会导致桥梁结构模态参数发生改变，从而导致基于动力特性测试的损伤诊断出现误判等问题，对江苏润扬大桥悬索桥的模态频率与环境条件长期监测数据进行处理，采用改进BP神经网络构建实测模态频率与温度、风速及车辆荷载的相关性模型。在此基础上分离环境条件变化对频率的影响，并采用假设检验的方法提出结构损伤预警方法。研究结果表明：基于提前停止技术和贝叶斯正则化技术的改进BP神经网络模型具有良好的泛化能力，可以有效地消除模态频率的环境变异性；采用t检验的方法识别出江苏润扬大桥悬索桥第5阶和第6阶频率的0.16%和0.12%异常变化，具有较强的损伤敏感性，适用于悬索桥结构的在线状态监测和预警。

英文摘要：

Considering that the variations of environmental conditions can lead to the change of modal parameters of bridge structures, which may cause misjudgment to the damage diagnosis by using the dynamic characteristic test, the long-term monitoring data of Runyang Suspension Bridge in Jiangsu Province was processed to establish the correlation model between modal frequencies and environmental conditions including wind velocity, temperature and vehicle loading by using the improved back-propagation neural network（BPNN）.Based on the model, the effects of environmental conditions on the modal frequencies were separated. Then the damage alarming method was presented by means of the hypothesis tests. The results reveal that BPNN-based model improved by early stopping and Bayesian regularization techniques exhibits excellent generalization capability, and the developed correlation model can effectively reduce the environmental effects in modal frequencies. The t-test method provides a good capability for detecting the damage-induced 0.16% and 0.12% abnormal changes of the 5th and 6th modal frequencies, respectively. Hence, the proposed method is suitable for real-time condition monitoring of suspension bridges.