Construction Project Cost Estimation Method Research Based On The RBF Artificial Neural Network

In recent years, with the smooth development of national economy and the demand of industrial transformation and upgrading, the real estate industry also changed its momentum of the explosive growth, gradually stepped into the stage of gradual development and industry consolidation. Most real estate enterprises start to give priority to the cost control and try to reduce the cost to ensure the profit maximization which has gradually become the mainstream.In the past time, the enterprises always rely on the construction drawing budget and completion settlement to calculate the cost of construction project, and in the construction stage, strict control of engineering change and replacement of engineering material was ailways happened in order to reduce the cost, the use of the favourable clauses in the contract to reduce the claim, all above might have certain effect of cost control, but its influence for the whole project is very low. Different from the traditional construction project cost estimation, now real estate enterprises has not only provide theoretical support for the feasibility study, but also has become an important way and means to build a dynamic cost index control system. Constructing a pecision cost index system can make the passive control to active beforehand control, which will directly affect the enterprises’ subsequent cost control work, and will be vital to the survival of the enterprise. This paper tries to use the method of artificial neural network to estimate index system of construction project cost.This article first expounds the research status quo of construction project cost estimation, and then introduces the basic theory of traditional construction project cost estimating. Then this paper introduces the basic concepts and principles of artificial neural network, and the principle and algorithm of RBF neural network has made the detailed description. Through the research of actual project data,put the Design-limited indicators as input variables, a long-term construction project can lead to a lot of uncertainty and risk factors,such as material price float, visa, claims, artificial floating wages and project progress control, etc.This things can lead the final actual cost and design-limited indicators is there a difference, different construction project risk is different, so the final difference is also different. Therefore design-limited indicators and construction of actual cost is not a simple linear corresponding relationship of the amount multiplied by the price,so we put the construction of actual cost as the the output variable, and establish the sample training set, looking for their nonlinear mapping relationship. We identified a large number of design-limited indicators through the actual project, using the principal component analysis screening of these indicators, eliminate the correlation between variables and reduced the number of input variables. These input-output indicators samples of RBF artificial neural network training set, Using the MATLAB software,set up the RBF artificial neural network model, And the training sample set and simulation,In the end is analyzed.