Application Of Neural Network In The Risk Of General Contracting Project

With the multiplicity and complexity of the projects, more and more building owners hope that contractors can supply general contracting service. General contracting projects applying fixed contracted price makes constructors be confronted with huge risk. Constructors must identify, analyse, appraise and take reasonable measures to decide the rate of risk compensation, but the estimation of the rate of risk compensation effects directly tender rates and profits of constructors. Risk of general constructing projects has the characters of uncertainty and dependency.In this thesis, through the discusses of characters, models and risk management of general constructing, we will adopt the analytical hierarchy process (AHP) methods to weigh the factors which affect the risk compensation rates with the example of "engineer-procure-construct" mode adopted by constructor, and the twelve key factors are determined. Based on the study of neural network, the radial basis function(RBF) neural network is constructed by Matlab6.X tools, which uses the key factors as inputs and the risk compensation rate as output, and then the network model is applied to analyze the sensibility and dependence of the key factors. Twenty specimens collected are used to training the RBF neural network, the others are used to test model. We draw the conclusion that1. The AHP is carried out to determine the key factors, which are used as inputs of radialbasis function neural network through numeralization, the nonlinear mapping from key factors to risk compensation rate is obtained through training neural network.2. Compare with other neural networks, the RBF neural network have the features of training quickly and little errors in estimating risk compensation rate.3. With the help of RBF neural network constructed, the sensitivity analysis of keyfactors and dependence between them are completed.