Analysis Of Data Mining Of Conventional Bridge Detection

Relying on the bridge intelligent detection system developed by China Merchants Highway Engineering Detection Center Co.,Ltd.,aiming at the more than 12000 bridges accumulated in the system,using statistical analysis,cluster analysis,association analysis and other methods,the bridge detection data are mined and analyzed,and the following main research results are obtained:(1)After preprocessing the data,we get a lot of valuable conclusions through multidimensional and multi-level statistical analysis.For example,in 12491 bridges,the number of girder bridges and arch bridges is 10140,accounting for 81.2%;the number of class 4 and class 5 bridges is 1988,accounting for 15.9%,indicating that the technical condition of the bridge is not optimistic and should be paid great attention to;the number of class 4 and class 5 bridges increases with the increase of operation time,with a rapid growth after more than 20 years of operation;the first six conventional bridges The major diseases in turn are: crack,peeling off angle,cavity and hole,honeycomb and pitted surface,concrete carbonization and steel corrosion.(2)In order to reduce the evaluation deviation among different inspectors,K-means algorithm and python program are used to cluster analysis and re evaluate the technical status of bridges.The results show that the proportion of 4 and 5 types of bridges is12.17%,3.73% lower than the original 15.9%.(3)Apriori algorithm and python program are used to analyze the relationship between the diseases of more than 12000 bridges.It is found that the probability of spalling and angle dropping is 60% when cracks occur,and the probability of cavity and hole spalling and angle dropping is 56%.(4)Taking a prestressed concrete beam bridge as an example,before the application of big data analysis,according to the inspection specifications and experience,the top five diseases of the upper bearing components of the bridge are successively the thickness of the protective layer,cavities,holes and peeling off,angle dropping,etc.After the application of big data analysis results,the top five diseases of the component are concrete carbonation,peeling,angle dropping,cracks(continuous beams),cavities,etc.,with the occurrence probability of 92.80%,85.6%,55.6%,29.7% and 10.0% respectively,which are more consistent with the diseases detected by the bridge.