Study On The Performance Evaluation Of Asphalt Pavement In XinJiang Region

Highway transportation is the basis for national economy development and a very important symbol of a country's level of modernization.Pavement performance is an important part which embodies the service level of highway, the evaluation of pavement performance is one key to understand the pavement condition and a sound basis to construct reasonable maintenance plan, in the mean time, it is the key to establish and the realize of pavement management system.This paper focuses on the evaluation methods of asphalt pavement performance, analyzes each individual index for evaluation of asphalt pavement performance, and according to the characteristics of asphalt pavement in the Xinjiang region, part of the individual indexes have been improved and perfected. Based on the individual index, the paper discusses comprehensive assessment methods of various asphalt pavement performance at home and abroad, element model and genetic neural network were selected as comprehensive evaluation methods of asphalt pavement performance in Xinjiang region.With the evaluation of element model method, the paper adopts piecewise function to construct correlation function, and analyzes the reasons and feasibilities. By the contrast of element model and PQI adopted by the technical specification of asphalt pavement maintenance, the shortcomings of specification methods and the advantage of element model were analyzed.With the evaluation of the genetic neural network model, two major branches of artificial intelligence:neural network and genetic algorithm were firstly analyzed.Aimed at the shortcomings of BP neural network, which is easy falling into local optimum, the paper proposes the introduction of genetic algorithm into neural network, using the complementary advantages of two, and combine the two organically to establish a genetic neural network model. Verified by measured data, the model is simple, efficient, and very suitable for evaluation of pavement performance with its powerful nonlinear mapping ability. so the paper ultimately determines it as the comprehensive evaluation method of asphalt pavement performance in Xinjiang region.On the basis of genetic neural network model, the paper analyses the evaluation results of asphalt pavement performance of different structure types and determines typical pavement structures which are suitable for different regions and traffic levels in Xinjiang.