Study On Forecast Model Of Operation Cost Of Urban Rail Project

Urban rail has become a lot of the city's main traffic way,the future will be faster to cover more of the city,but urban rail transit operation cost is too high,the government in the project feasibility study report forecast operation cost,only use a simple linear relationship between the forecast of expenses,only consider the traffic volume of a single variable,overlooked some nonlinear relation between cost and operation workload,and these expenses accounts for larger proportion of operating costs,lead to urban rail project actual operating costs and forecast cost difference is too big,so more accurate projections for urban rail project cost,It is important to provide a basis for government decision-making.This article first analyzes the urban rail project operating costs,as well as the factors that influence the operation found that operating costs is affected by a variety of operating workload,cost data is nonlinear relationship again,the cost of traditional prediction method ignores the characteristics of urban rail operation cost,so this article combined with the characteristics of urban rail operation cost,use the homework cost method and BP artificial neural network to predict operational costs of the new urban rail project.This paper according to the operation of urban rail operation project activity type operations can be divided into two processes of operation,the overhaul,according to homework cost method could be divided into a number of assignments,activities will be operating costs accrue to the operation center,choose the operating cost drivers,using correlation analysis,cluster analysis and principal component analysis combined with relevant cost drivers,determine the final cost drivers used to predict operating costs;Second,prediction using BP artificial neural network fault tolerance,the nonlinear mapping ability,to the urban rail operation enterprise operating costs and cost drivers for learning training calendar year,the preprocessing of sample data,build the network structure,learn parameters,established between the operating cost and cost drivers of BP artificial neural network model,the accuracy of the test model with test data,the formation of trained prediction model;Thirdly,the accuracy of BP artificial neural network model prediction is determined by comparing the prediction results of urban rail project operation cost with the prediction results of multiple nonlinear forecasting methods.Finally,the single-factor sensitivity analysis method is used to analyze the sensitivity of key cost drivers and determine the cost drivers that focus on.