The Research Of Uncertain Project Estimation's New Method Based On RBF Neural Network

Construction cost includes investment estimation, design budgetary estimate and construction budget, on which to control the investment in different phase. Investment estimation is an important content of project's available research, and also is the basis of design budgetary estimate. So its accuracy will affect directly that whether the project can include in the future destiny of the construction plans. Now our country's cost estimation is still based on estimation index and the ration. With the further development of our socialist market economy, the cost estimation method in our country will transfer to the method of the western country in the end, and separated from the "quality" and "price" becomes an urgent request. Being the foundation of the neural network method applying for investment estimation, the method based on neural network has certain meaning. In the research of investment estimation,back-propagation neural network has been used into construction engineering, but still belongs to the starting stage. However for adopting a gradient-descent approach of adjusting the neural network weights, back-propagation training has some drawbacks that influence its practicality. This paper proposes a superior feed forward network called radial basis function neural network (RBFNN), proposes uncertain projects' investment estimation based on neural network. It is presents a novel and effective method for project investment estimation.