Application of Data Cleaning and Early Warning Strategy in Bridge Health Monitoring System
Bridge health monitoring data is difficult to serve the practical safety operation of bridges.A significant portion of this issue arises due to data distortion and inadequate warning strategies.In order to solve this problem,this paper analyzes and summarizes the common data distortion types and recognition and repair algorithms at home and abroad,establishes a data cleaning strategy,and formu-lates a damage recognition and early warning technology for bridge finite element model based on multi-layer neural network by using the multi-level coupling method of data-driven and finite element modeling.According to the comparison and analysis of the histori-cal monitoring data of the health monitoring system of Rugao Jiaogang Bridge,the accuracy and sensitivity of the data cleaning strategy are more than95%,and the three-level early warning system established is in line with the change law of the original strain,which solves the problem of false alarm caused by the unreasonable threshold setting of the system.
bridge health monitoringdata cleaningearly warning
NETL
NSTL
万方数据
