Research On Damage Early Warning Method Of Platform Canopy Based On Structural Health Monitoring

With the rapid development of high-speed railways and the new requirements for the unique shape of various types of platform building facilities,platform canopies usually have the characteristics of complex structural forms,long-term durability design,and are not easy to replace.Today,platform canopy safety accidents occur frequently and are discovered in time.It has become necessary to realize damage warning in damage conditions.It is of great significance to explore and improve the early warning methods of structural damage in the field of railway house construction for the development of platform canopy structural health monitoring systems.Based on the structural dynamic response acceleration parameters obtained by the structural health monitoring system,based on the engineering background of the platform canopy structural health monitoring system,this paper studies the damage identification and early warning methods for the symmetrical structure of the platform canopy adjacent railway lines in the railway field.The damage early warning method based on the time series similarity measure FED distance function index and the PED and Hotelling T~2damage early warning methods based on principal component analysis were carried out.The FED-Hotelling T~2 damage early warning method combined with FED was proposed.The main research contents are as follows:(1)Based on the analysis of actual engineering conditions,a structural health monitoring system is established to collect and analyze the structural health monitoring data,and data preprocessing methods such as data noise smoothing,abnormal data removal,and default value completion are studied.Investigate the processing method of the subject.(2)Analyzed the similarity influencing factors and similarity measurement methods of time series,combined with the actual situation,studied the feature extraction of time series data based on time series frequency domain conversion,and selected a time-frequency domain conversion method suitable for the research subject in this paper.The distance function index of Euclidean distance and cosine similarity was analyzed,and the applicability and damage sensitivity analysis of the two were performed by numerical simulation methods.It was verified that Euclidean distance as a warning indicator is sensitive to the degree of damage and not adjacent to the damage site.Insensitive features.(3)The principal component analysis based on data mining is researched,and the PED distance function indicator based on principal component analysis is used to predict the overall damage of the adjacent structure of the platform canopy,and the Hotelling method is proposed based on the principal component analysis.T~2 damage warning method.The validity of the method is verified by numerical simulation and analysis of actual monitoring data.(4)Comparing the applicability of the two methods,a damage warning method combining FED and Hotelling T~2 methods is proposed,and the method of selecting the warning threshold is summarized.The results show that the method is feasible.The damage warning method combined with FED-Hotelling T~2 method presented in this paper shows high recognition accuracy in the damage warning of the symmetrical structure of the adjacent platform of the platform canopy,which is in line with the online data collection,data preprocessing,and The similarity metric distance function index is directly related to the construction of the Hotlin statistic and can be used in the early warning research of the damage of the adjacent structure of the platform canopy.