基于神经网络的预应力混凝土连续梁桥健康监测数据拟合研究

Research on Health Monitoring Data Fitting of Prestressed Concrete Continuous BeamBridge Based on Neural Network

某互通匝道预应力混凝土连续梁桥支座位移及桥墩倾斜在线监测为工程背景,基于既有监测数据建立的神经网络,建立各个测点支座位移、桥墩倾斜及温度之间的关系模型,并基于所建立的模型拟合数据,分析结果表明,基于神经网络建立的分析模型所拟合的结果与实测值较为吻合,满足工程需要,所拟合的数据可以对实际工程中由于现场环境、传输问题及硬件故障等原因导致的缺失数据进行重新构造,以减少数据丢失对结构监测以及安全评估的影响,并为运营人员管理及研究人员分析问题提供帮助,本文的研究方法可为类似工程提供参考及指导。

Based on the online monitoring of support displacement and pier tilt of a prestressed concrete continuous beam bridge on an interconnecting ramp,and the neural network established by the existing monitoring data,the relationship model among support displacement,pier tilt and temperature of each measurement point is established,and the data are fitted based on the established model.The analysis result shows that the fitting results of the analysis model established based on the neural network are consistent with the measured values and satisfy engineering needs.The fitted data can reconstruct the missing data caused by the field environment,transmission problems,hardware failures and other reasons in the actual project,so as to reduce the impact of data loss on the structure monitoring and safety assessment,and provide help for operator management and researcher analysis.The research method provided in this paper can provide reference and guidance for similar projects.