Evaluation And Prediction Of Freeway Semi-rigid Base Asphalt Pavement Performance Base On Support Vector Machine

With the increase of service and traffic,freeway in China gradually change from the stage of rapid construction to the stage of maintenance and management.The task is very arduous,and evaluation and prediction of pavement performance is the basic of pavement maintenance and management.Because the main structure of freeway in our country is the semi-rigid base asphalt pavement,this paper systematically studies evaluation and prediction of freeway semi-rigid base asphalt pavement performance.Based on the damage data of Luo-ning freeway of Fujian in 2016,Beijing-Tibet freeway(Section in Gansu)in 2016 and 14 freeways of Hebei in 2014,this paper analyzes the main damage types of freeway semi-rigid base asphalt pavement,then the index system of pavement performance is established.This system includes three types of indices,including pavement surface condition index,pavement structure strength index and pavement function performance index.The pavement surface condition index includes three secondary indexes,they are pavement patching ratio,pavement cracking ratio and pavement distress ratio.The pavement function performance index includes riding quality index,rutting depth index and skidding resistance index.Based on Support Vector Machine,this paper builds two models that can evaluate and predict the freeway semi-rigid base asphalt pavement performance.The evaluation index is reduced by Principal Component Analysis Method to form independent principle components,then select the principle components samples to train Support Vector Classification to build PCA-SVM model combining Principal Component Analysis and Support Vector Classification,the results of the model are consistent with the actual situation.The parameters of Support Vector Regression is optimized by using the strong global search ability of Genetic Algorithm to solve the problem that it is difficult to select accurately,and the GA-SVR model combining Genetic Algorithm and Support Vector Regression is built,the relative error of the prediction results is less than 1%.The research results are used to guide the accomplishment of the maintenance plan and scheme of freeway semi-rigid base asphalt pavement,and provide reliable theoretical basis for maintenance decision and pavement management.