Early-warning Method Of Bridges Anomaly Monitoring Data Based On Time-frequency Analysis

Bridge structural health monitoring is an important guarantee for bridge safety.Monitoring data early warning is an important task in the field of structural health monitoring.At present,the two key issues of monitoring data early warning are real-time and accuracy.In order to solve these two key problems,based on the data of a high-speed railway bridge,this paper analyzes the characteristics of the monitoring data and studies the early warning of the monitoring data in both time and frequency domains.The main research work of this article is as follows:(1)The data collected by the vehicle speed sensor often has errors and cannot be used directly.In response to this problem,this article starts from the Doppler effect and the basic principles of the Doppler radar speedometer to study the root cause of errors in vehicle speed data.The vehicle speed data and related vibration sensor data are analyzed and observed,and the time error and space error of the vehicle speed data are summarized and refined.Aiming at the spatio-temporal errors of vehicle speed monitoring data,an automatic recognition algorithm for vehicle speed considering the spatio-temporal errors of monitoring data is proposed.Comprehensive analysis is carried out through a variety of error recognition methods to accurately extract the vehicle speed data segment and then accurate vehicle speed data is calculated.(2)At present,the abnormal analysis of the monitoring data is mainly based on offline batch processing,and it is impossible to analyze the data in real time online.So there is a certain lag in the abnormality detected,and the advantages of the monitoring system cannot be used.To solve this problem,anomaly monitoring based on hierarchical temporal memory is adopted.The hierarchical structure,sparse distributed representations,and learning prediction rules of this method make it have the characteristics of high data utilization rate,high efficiency and strong robustness.The anomaly in the monitoring data is analyzed,and has the characteristics of sensitivity and efficiency.(3)The forced vibration data under train load is easy to be misreported as abnormal due to its large amplitude.Based on this problem,the structure of the train was analyzed,and the vibration response of the bridge was theoretically deduced.The spectrum characteristics of the vibration response were studied.The analysis shows that the vibration response monitoring data under the train load has a certain periodicity,which is mainly based on the forced vibration caused by the train load The vibration response monitoring data of the train during the opening phase will have multiple obvious peaks from the power spectral density,and the frequency corresponding to each peak is a forced vibration frequency in a multiple relationship.The moving window spectrum analysis is proposed,and the appropriate parameters are selected for different data.This method can accurately identify the monitoring data under the load of the train and has a strong timeliness.(4)A method for early warning of abnormal vibration monitoring data of bridges based on time-frequency analysis is proposed.This method combines the advantages of hierarchical temporal memory and moving window spectrum analysis,and makes up for the deficiencies of the two methods.The results of early warning of abnormal vibration monitoring data of bridges based on time-frequency analysis are summarized into three types of normal data and three types of abnormal data.The results of numerical examples show that the method can effectively distinguish normal data from abnormal data,and verify the effectiveness of the abnormal warning method of bridge vibration monitoring data based on time-frequency analysis.