【摘要】 为了充分利用桥梁健康监测系统采集的海量数据,健全大跨钢箱梁斜拉桥结构健康评估方法,以辽河特大桥健康监测系统为依托,运用小波分析法对辽河特大桥南北塔为期10个月的支座位移数据进行了滤波与重构.通过大量试验选取神经网络参数,构建了BP神经网络,对重构后的支座位移与环境温度相关性模型进行训练,建立人工神经网络评估模型,并对模型进行检验,检验结果表明:对神经网络预测值与实际值误差取5%的显著性水平作为误差代表值能够有效地反映结构健康状况.更多还原

【Abstract】 In order to make full use of the massive data collected by bridge health monitoring system and improve the large-span steel box girder cable-stayed bridge structure health assessment method,based on Liaohe super large bridge’s health monitoring system,the south and north tower pedestal displacement data of the bridge are filtered and reconstructed for a period of 10 months with wavelet analysis method.Meanwhile through a large number of tests,the parameters of the neural network are selected and BP neural network is established.The pedestal displacement and temperature correlation model is trained after reconstruction,and artificial neural network evaluation model is established and tested.The results show that:selecting 5% of the significance level error as the representative value of the neural network forecasting value and the actual value can effectively reflect the health condition of the structure.更多还原