Research On The Prediction Methods Of Rational Scale Of Regional Highway Network

Highway is the important infrastructure of national economy and defense, its developing scale is an important signal of one country or region. Convenient and fast highway network can not only make people live conveniently but also promote economy. The construction of highway must meet national economy and society development or be advanced, during the different stage of the society, highway should have relevant scale. So, research on the rational scale of regional highway network, holding adaptation between road traffic and economy, finding out problems during the construction of the highway network, defining developing train,target and focal point, has important significances about promoting road traffic and economy develop harmoniously, achieving nowadays and sustainable development of road traffic.But we can find out that highway construction costs huge and funds are hard up seriously. We must research rational scale of regional highway in order to improving economic benefit of highway construction investment when funds are limited. If highway scale exceeds real traffic demands, it will lead construction funds to being engrossed prematurity, and bring out land wasting and much cost for dismantle, which is opposite to the short construction funds situation of our country; when the scale is over low, the programming highway after being used will bring out results such as traffic saturation,low transportation efficiency and traffic accident happening frequently etc. because of heavy traffic, and will restrict regional economy. So defining the whole scale based on economy level, and modulating investment intension has important significance for settling problem between construction demands and short funds, guaranteeing highway traffic and national economy develop continually.The paper uses system analysis to set up in order arrangement construction of influences about highway network based on relation between economy and highway traffic, meanwhile according on developing progress in developed country, concludes increasing rules of regional highway, analyses that highway construction goes through arising,developing and mature these three stage. Addressing that highway construction has few typical characters such as over economy,several phases and flexible quotiety attenuation particularly. These settle bases for further research.Analyses several general prediction methods of national scale of regional highway network, aiming at oneness,unilateralism of general prediction methods, puts forward combination prediction methods. For combination prediction methods, what is important is confirming weighing of single prediction model, general linear combination may come out minus weighing, which is illegitimate. Using linear combination model can not guarantee precision because influences about highway network is all sides and has complex interrelation. BP neural network has high learning ability,adapting ability and nonlinear ability, so the paper builds nonlinear combination prediction model based on BP neural network. Choosing country land quotiety method,factor analysis method and growth curve model as inputs of network, real scale of highway network as output, does training. When net is mature, satisfies precision request, then predicts the scale of highway network.The paper confirms least cost,passing ability most and technical level highest to be the aim of optimization of grade construction, improves existing multiobjective planning model, furtherly builds multiobjective planning model of highway network technical level optimization., and applies this model to find out scale of highway network technical level optimization in JiLin province, providing theoretic basis for highway traffic planning in JiLin province.Finally, finds out problems after analysis of developing conditions for highway network in JiLin province. Applying country land quotiety method,factor analysis method and growth curve model to does prediction , then using these three results as inputs nods of network, corresponding real highway length of each year during 1990-2005 as outputs, builds 3—20—1 neural network structure, using Fletcher-Reeves conjugate gradient algorithm because it has rapid speed and less memory. Achieving this by training function traincgf( ) in MATLAB. Then predicts the scale of highway network of year 2020 in JiLin province. This model considers all factors'influences and rules of highway development selfly, after demonstration, proves rationality and veracity. Using multiobjective planning model to define highway network technical level optimization in JiLin province, it has definite theoretic and scalar directing sense for highway traffic development in JiLin province.