Research On Damage Alarming Of Railway Steel Bridge Based On Analysis Of Health Monitoring Data

Bridge construction has made remarkable achievements along with the rapid development of railway especially high-speed railway.However,due to the increases of the bridge span,the increasing complexity of structure type,the influences of uncertain elements,furthermore the structure extended active duty,ultra limit of service,corrosion,fatigue,even the occurrence of hazardous events,for instance the sudden earthquake,the collision of vehicle,ship and so on,safety accidents of bridge structure occur frequently,which requires that the bridge health monitoring system can detect the occurrence of damage promptly and give early warning in time.In order to ensure the safety of bridge structure,comprehend and master the health status of the bridge in real time,and avoid the occurrence of safety accidents,bridge health monitoring system has been installed in many bridges at home and abroad,but there are few studies on damage warning methods based on monitoring data analysis,it is of great significance to study and improve the early warning method of bridge damage for the development of bridge health monitoring systems.Based on the background of health monitoring system for railway steel bridge project,circumfuse the method of damage alarming,by means of time series similarity distance function index based on domain transformation feature representation and the AR coefficient of the ARMA model,damage alarming has been studied.The main research contents and conclusions are as follows:(1)Combining the characteristics of bridge health monitoring data,the process of monitoring data preprocessing is summarized,the data processing methods of missing data,abnormal data,trend and noise data are studied,the processing criteria and methods suitable for this article are selected,wrote programs to implement the process and verify the effectiveness of the methods.,according to the health monitoring data of the bridge,the train traffic safety and the state evaluation based on the frequency index are analyzed.(2)The methods of measuring the similarity of time series and the theory of vibration signal domain transform frequency division are summarized,a damage warning index of measuring similarity by distance function is proposed,including the Pearson correlation coefficient,cosine similarity and Euclidean distance,based on feature representation of the original monitoring time series data through vibration signal domain transform frequency division,numerical simulation of responses of moving loads of simply supported steel trussed girders under different damage conditions,used distance function to carry out damage alarming,validated the effectiveness of the distance function as an early warning indicator.(3)Based on the health monitoring system of single-line simply supported steel truss bridge and multi-line continuous steel arch bridge,summarize the layout characteristics of the measuring point,select suitable measuring points and working conditions,carry out damage warning using distance function of time series similarity based on domain transformation feature representation as an indicator,the result of early warning indicates that it is feasible to use distance function index to carry out damage alarming by actual data analysis,summed up warning threshold determination method,selected damage warning threshold of distance function.(4)A time-domain ARMA model for acceleration data of the monitoring system was established by programming procedures,the reconfiguration and predictability of the model are verified,the law of model coefficient of different acceleration data is analyzed,the sensitivity of the model coefficient to damage is verified by numerical simulation,the results show:The first three-order AR coefficients of the model can well characterize the inherent characteristics of the structure,furthermore they are more sensitive to damage.A damage warning index based on AR coefficient is proposed,and it's validity is verified through a numerical model.Summarized warning process of measured data,damage warning for actual bridge was conducted,based on the data of health monitoring system.