Bridge Structural Health Condition Recognition Based On Time Series Data Analysis

With the development of society,the traffic increasing load requires more efficient bridge service.Therefore,the technology of bridge health monitoring system has been developed and widely used.It is an important research topic to analyze bridge monitoring data and to identify the bridge structural health status efficiently and accurately.Based on the analysis of time series data,this paper aims to identify the bridge structural health status in real time,and take timely measures to repairing bridges with high degree damage in case of serious traffic accidents.The key to identify bridge structural status is to accurately classify the bridge monitoring signals of in real time.The data obtained directly by various sensors in bridge structural health monitoring system are typical time series data,with migration,compression,or stretching on the time axis.Therefore,the traditional health status recognition methods are suffered in real applications.According to the problems that there is the phenomenon of migration,compression or stretching of bridge structure monitoring data even under the same structure health state,this paper firstly uses dynamic time wrap algorithm to align the deformable time series,and then uses the classification algorithm based on k-Nearest Neighbor to predict the bridge structural health state to solve the problem of classification accuracy degradation intraditional distance algorithm which cannot handle the deformable time series.In addition,the time series data obtained by monitoring are generally huge sized.And it is difficult to quickly obtain the classification results with high precision using common algorithms.In this paper,a Long Short-Term Memory network is proposed to deal with the noisy and verbose bridge structure monitoring time series data.Afterwards,we use the network trainded by a certain amount of data to identify the health state of bridge structure quickly and accurately.By accurately predicting the health state of bridge structure,this paper makes us more understanding of the monitoring time series data of bridge structure,and approaching the real-time analysis of the original data,which will greatly improve the efficiency of the bridge structural health status identification.This paper uses the real bridge data provided by the structural monitoring and control research center of Harbin University to verify the effectiveness of the algorithm for analyzing bridge structural monitoring data.By further comparison experiments,it is proved that the proposed algorithm is more applicable to practical applications.