Research On Evaluation And Decay Prediction Method Of Expressway Pavement Performance

Expressway is an important part of transportation in our country.In the early stage of large-scale construction of expressway in China,a considerable part of expressway sections have entered the stage of major and medium-sized repairs.This puts forward new and higher requirements for the whole process of highway pavement performance detection,evaluation,prediction and maintenance.At present,the evaluation model of expressway pavement performance is too subjective and absolute.There are some problems in pavement performance decay prediction,such as large errors and some useful information being discarded.The main research content of this paper is to establish a more objective and comprehensive comprehensive comprehensive evaluation model of pavement performance and prediction model of pavement performance decay based on the existing data,using the idea of combination,in order to provide more accurate maintenance decision-making basis for highway pavement maintenance management departments.Firstly,this paper introduces the types of highway pavement structure and damage types of different pavement structures in China and analyses their causes.At the same time,the evaluation methods and standards of individual evaluation indexes for expressway pavement performance in different specifications are compared and analyzed.Finally,two different combination methods are used to establish the combined prediction model of highway pavement performance degradation based on the combination principle.The BP neural network and exponential smoothing method are used in the single prediction model.According to the characteristics of different membership functions,ridge distribution membership function and trapezoid membership function are selected as membership function of fuzzy comprehensive evaluation of expressway pavement performance,and a fuzzy evaluation model of single performance evaluation index is established.In order to improve the reliability of model evaluation,a fuzzy comprehensive evaluation model of combined membership function was established by means of mean square deviation weighting method.From the example,it can be seen that the combined membership function fuzzy comprehensive evaluation model can effectively reduce the fluctuation of comprehensive evaluation of pavement performance,make the comprehensive evaluation results more stable,and solve the problem of absolute evaluation results when the membership degree is similar.Among them,in the comprehensive fuzzy evaluation model,the weight of single evaluation index of pavement performance is determined by G1 method.This method can meet the needs of maintenance departments for different maintenance points of road sections,and it is a better method to determine the weight of single evaluation index.Finally,a combined forecasting model for pavement performance degradation of expressway based on two different combinations is established.After discussing the influencing factors of pavement performance,the existing prediction models of pavement performance and the related theoretical basis of combination prediction,in order to reduce the prediction error of pavement performance,the single prediction model of BP neural network is established by choosing road age and traffic volume as influencing factors,and the smoothness value of exponential smoothing method is regarded as the expression of climate-affected factors.Subsequently,the combination forecasting model based on the entropy weight method and the generalized average combination method is established,and the MATLAB program of BP neural network single prediction model and generalized average combination prediction model is compiled,and the MATLAB program of parameter P solution when the minimum error of generalized average combination prediction model is obtained.Through the case study,the combination forecasting method can effectively reduce the prediction error of pavement performance,and the generalized average combination method obtained by solving the minimum error model parameter P has smaller prediction error and it is a better combination forecasting method.