Research On Railway Bridge Damage Identification Based On Time Series Similarity Measure

With the increase of the service time of the bridge,under the influence of the external environment and traffic load,the problems of component failure and material aging also follow.In order to grasp the operating status of the bridge in time and identify the damage of the bridge,many bridges are A structural health monitoring system was installed.However,with the operation of these structural health monitoring systems,the massive amount of collected data is backlogged in the database and has not been effectively used.Therefore,how to effectively use these massive bridge health monitoring data to extract information related to the state of the bridge structure,It is of great significance to identify bridge damage.Based on the bridge health monitoring system of Nanjing Dashengguan Bridge on Beijing-Shanghai High-speed Railway,this paper conducts the research on bridge damage identification from the perspective of the correlation of multi-point data and the similarity of single-point data.The main research contents of this paper are as follows:(1)Based on the measured bridge health monitoring data of Nanjing Dashengguan Bridge,the basic process and processing methods of data preprocessing are studied;then the components of train loading frequency are extracted from the original vibration response of the bridge,which represents the impact of train load on the bridge.And based on this component,the cross-correlation between the acceleration measurement points of the two main spans of the Dashengguan Bridge is analyzed;finally based on the above analysis,the cross-correlation diagram of the root mean square value and peak value of the vertical acceleration of the two main spans Performed linear fitting and established a control chart of fitting coefficients to identify the damage of the bridge by the changes of the fitting coefficients before and after the damage of the bridge;(2)The time series similarity and its measurement method are studied,the dynamic time warping algorithm in the field of speech recognition is introduced,and the basic process of measuring time series similarity is introduced;then the method is applied to the damage identification of bridges.: In the same direction,when the same type of train passes the same bridge at the same or similar speed,the bridge vibration response before and after the bridge damage is not similar;based on this principle,the above dynamic time warping algorithm is used to respond to the vibration acceleration of the two bridges The time series are similarly measured,the DTW distance between the two series is calculated,and the damage warning index RDS is set based on the calculation results;finally the bridge finite element model is established,and different bridge damage conditions are set,which is effective for the proposed method Has been verified;(3)Using the proposed bridge damage identification method based on the dynamic time warping algorithm,a damage warning study was carried out on the acceleration response data of the main span of the Nanjing Dashengguan Yangtze River Bridge,and the correlation between temperature and RDS indicators was analyzed;The health monitoring data of the bridge is divided into training samples and test samples.Based on the distribution of the maximum daily RDS of the training samples,the yellow warning threshold and red warning threshold of the RDS indicator are set;finally,the acceleration data is increased in different degrees to simulate The damage of the bridge verifies the effectiveness of the method.