Bridge Deterioration Analysis Based On Censored Data

Bridge deterioration reflects the change of bridge condition.Due to the influence of internal and external factors,bridge deterioration is an inevitable process of bridge life cycle.It refers to the defect of bridge components,which gradually affects the function of the bridge.Accurate bridge deterioration analysis can provide a more scientific basis for making plan of bridge maintenance,reinforcement and reconstruction,which can ensure the operation safety of highway bridge.Bridge deterioration analysis is to establish a prediction model according to the influence degree of bridge deterioration factors on the bridge condition grade or bridge score,and to predict and analyze the future condition of bridge and deterioration factors.Many methods such as regression analysis,Markov chain and fuzzy technology have been applied to bridge deterioration analysis and achieved some results,but the existing analysis methods still have the following two defects.First,nature environment and climate factors are not considered.For bridges that is located in the area with special geographical environment and climate,the factor analysis of bridge deterioration is not comprehensive enough.Second,they neglected the existence of censored data.Censored data is extremely common in data sample of bridge degradation analysis.It means that data is truncated due to loss of follow-up and withdrawal of the bridge during the observation period,or the bridge did not deteriorate at the end of the observation period,as well as the limitation of the scope of the bridge scoring standard on the correct evaluation of the bridge,which results in incomplete data.Therefore,there may still be deviation in the prediction results of existing analysis methods,making the existing analysis methods unsuitable for the actual bridge management and maintenance.In view of the above problems,this paper proposes a bridge deterioration analysis method based on the censored data.This method adds nature environment and climate factors into deterioration analysis,and the corresponding attributes are added in the data set.At the same time,this paper discusses the type of censored data in the bridge data and the reason for censoring based on the idea of survival analysis,and adopts the method based on Cox regression model to process the censored data and construct the prediction model of bridge deterioration.In order to ensure the objectiveness of experiment,raw data collected from the Yunnan Communications Investment & Construction Group LTD are used in this study.The baseline method is the multiple linear regression analysis method,which is usually used in multivariate analysis.The experimental results show that the bridge deterioration analysis method based on the censored data proposed in this paper can correctly process the censored data compared with the traditional baseline method,and it has higher prediction performance and application value.Thus it verifies the effectiveness of the method proposed by this paper in bridge deterioration analysis based on censored data.