Research On Preventive Maintenance Of Asphalt Pavement In Seasonal Frozen Area Based On BP Neural Network

With the development of my country’s highway traffic,road maintenance has received more and more attention.As part of the maintenance method,preventive maintenance has attracted more and more attention because of its huge benefits.However,in practical applications,preventive maintenance decisions are often made based on the subjective experience of the constructors.At present,more research was made on the application technology of pavement preventive maintenance technology in maintenance measures,and less on preventive maintenance decision-making technology.Jilin Province is in seasonally area,and hot in summer,severe cold in winter which are extremely unfavorable for temperature-sensitive asphalt pavement.Asphalt pavement diseases here are various-kinds,earlier-occurred and quickly-developed.Therefore,the amount of money invested in asphalt pavement maintenance is huge.To prevent problems by taking preventive maintenance measures before occuring is the "key" to solve the above problems.The traditional artificial decision-made method has low efficiency and is greatly affected by subjective factors of the staff.The main research content of this paper is deciding preventive maintenance measures for highway in Jilin Province.On the base of domestic and foreign research literature,high diseases in more detail are classified in the seasonally frozen area from the geometric characteristics and appearence.For common highway diseases in the seasonally frozen area,four categories(cracks,loose,deformed,and other types)were classified.And each disease of mild cracks,potholes,rutting,subsidence,waves,and flooding was graded in terms of length,area,depth,performance characteristics and other aspects.Summarize of the preventive maintenance measures currently used for road diseases,including crack treatment methods,sand-containing mist seals,slurry seals,micro-surfaces,gravel seals,fiber seals,composite seals,and ultra-thin wear layers,Thin-layer overlays,etc.were introduced and compared.At present,preventive maintenance decision-making is affected by many factors from construction personnel.On the base of the comprehensive asphalt pavement diseases investigation in seasonally frozen areas,this paper quantifies them and establishes a "disease library".At the same time,it establishes asphalt in accordance with the maintenance specifications of seasonal frozen areas.The pavement "preventive maintenance measures library" uses the BP neural network in the MATLAB toolbox to realize the classification and gradation of diseases,and based on the result,the "disease library" and "maintenance measures library" are sampled for training and learning,and the corresponding mapping are established.The preventive maintenance model of highway diseases provides a new method for the decision of highway preventive maintenance,which avoids subjectivity of artificial method and can lower life-cycle cost.