| After nearly 40 years of development,bridge health monitoring technology has accumulated a large amount of monitoring data,but the analysis theory and analysis methods for these data have not yet been developed and perfected.The full use and effective mining of data have not been achieved,resulting in the shelving and waste of massive data,which can not give full play to its due value.It is the core of practical application of bridge monitoring that how to excavate reasonable data processing methods from the accumulated large amount of data,analyze and process mass monitoring data correctly and timely,and obtain effective information.Bridge bearings are important components connecting the upper and lower structures of the bridge,and are subject to the combined effects of environmental and vehicle loads during their operation.Once the performance of the bearings changes,the normal operation of the bridge will be seriously affected.Therefore,the study of bridge bearings is of great significance to the overall study of Bridges,and it is also a key link to achieve the state assessment and performance prediction of Bridges.The research on bridge bearings has become one of the more important bridge research topics.This article takes the temperature,vehicle and support displacement data of A long-span railway steel truss arch bridge as the research object,and mainly studies the influence of temperature and vehicle on the support displacement.Combined with the application of MATLAB software,a new method for establishing a mathematical model of temperature and support displacement is proposed,and a BP neural network model is established to perform classification extraction and statistical analysis of different types of response of support displacement under different driving directions of the train.(1)Investigate the current research status of bridge health monitoring data,indicating that the completion of feature extraction and mining modeling of monitoring data is still a relatively core part of current health monitoring systems,and summarize the existing research experience and methods to lay a foundation for the following research.(2)Summarize the different types of bridge bearings and their functions,and the damage mechanism of support is analyzed by taking plate rubber bearing,basin rubber bearing and spherical steel bearing as examples,and specific prevention measures are put forward according to the causes and mechanisms of support diseases.(3)Make good preparations for data mining modeling of a long-span railway steel truss arch bridge studied in this paper: the development of MATLAB's data sieving and pre-processing graphical user interface(GUI)and the development of the Figure graphical data selection function toolbox,And introduce the data mining modeling methods: the basic methods and principles of linear regression and artificial neural networks.(4)The relevancy between the support displacement and the environmental temperature,the support displacement and the temperature fields of the beam was analyzed,and the temperature of the five positions of the beam with strong linear regression and the two sets of temperature difference data were selected to establish principal component regression model of support displacement and temperature by using the principal component regression method.(5)Create a BP neural network model to classify and extract the displacement response of different types of bearings corresponding to trains passing in different directions.and select the two types of support displacement response types that best reflect the bearing sliding performance under different driving directions,the statistical analysis of the bivariate probability density distribution of the displacement value of the bearing and the change amount of the displacement of the bearing under the condition of gentle temperature characteristic changes is obtained.The threedimensional kernel density estimation graph can be used to evaluate the performance of the bearing. |
PDF Full Text Request