Research On Project Cost Estimation Model Based On The GRA And BP Neural Network

By the end of2014, China has rapidly become the first countries of highway totalmileage in the world. The rapid development of highway will inevitably need sustainedlarge investment, the project cost estimation as an important basis for the project prior tothe construction investment decision-making play a decisive role. However, the shortageof highway cost estimate is not fully reflect the nonlinear relationship between predictorsand unit mileage cost, and the weight of each estimate index is not considered enough,the method of determining weights is not convincing. Therefore, scientifically build theindex system of highway project cost estimation, and determine the weight of each index,uses a reasonable method to establish the highway construction cost estimation model isthe need of this dissertation to solve the problem.Firstly, this dissertation summarizes the project cost estimation research at homeand abroad, analyzes the problems existing in the highway project cost estimation in ourcountry; Then elaborates the relevant theory of highway and project cost, correlationtheory highway project cost estimation method in detail, lays the foundation for later costestimation; Then analysis the main project constitution of the highway engineeringproject cost, selects the appropriate project characteristic index, builds index system ofhighway project cost estimation; Finally, combines the grey relational analysis（GRA）and BP neural network to estimate highway cost, grey correlation analysis is used todetermine the predictors of weight, BP neural network is used to establish predictionmodel. Through the empirical study demonstrates that the accuracy of estimated resultsis high, indicates that the research method of this dissertation is feasible, and provides anefficient and reasonable method for China’s highway cost estimation in the future.