The Research Of The Model In The Estimation Of Construction Project Cost

There are a lot of problems in project investment control for a long time. For example, it always happens that budgetary is beyond basic estimate budget. This brings a lot of lost to the nation and the investor. Nowadays, more and more investors have realized the importance of the investment control. It is an important research to find a suit method to estimate the project cost exactly in the early period.This paper constructed one kind of cost estimating model based on the project construction cost database supporting, on the basis of nonlinear mapping relation between project characteristic and project construction cost, combining the artificial neural network's non-linear mapping ability and self-learning capacity with the fuzzy mathematics'pressing close degree, and apply to the construction project carry out example verification.In order to avoid subjective error in the artificial sample choice, the fuzzy mathematics'pressing close degree was used to choose training samples in this paper. Which can express qualitative description of project information in digits, make the process of training sample be artificial intelligence product, and make the large-scale sample selection in project costs databases by applying a computer become possibility, Based on above works, some research on indicators system of similar project, quantifying project characteristic factor,and data handling are made.To overcome the shortcomings of the standard BP neural networks in actual application, such as larger and more complex network structure, slower ANN convergence, and weak generalization capacity, this paper presents some solving methods. Grey relational grade analysis was applied to optimize network structure. Adding up inertia item and adaptive study rate method was conducted to improve the convergence rate of ANN. Neural network ensemble was proposed to improve the generalization capacity. Thus the entire model possess great application value.