Study On The Cost Prediction Method Of Building Based On RS-WNN

The high-activity predicting of the cost building is necessary to the practice and theory of project management. According to the present research of the cost building, in the thesis, an intelligent prediction system is composed of 4 mathematical methods, including Rough Sets(RS), unascertained measure, wavelet analysis and artificial neural networks (ANN), to solve the hard problem.In the thesis, the theory and method is introduced. At the same time, the character and merit of wavelet neural networks (WNN) is analyzed, including the learning rules and the procedure of training. WNN is better than NN in adaptability, learning speed and high-precision. In the thesis, the intelligent prediction system is used in the area of the cost building using the main framework of WNN.And we use RS-WNN system to predict the cost of building. There are many factors affecting the cost of building, it makes hard to train, if we use all the factors to be input nodes of networks, some of the factors are related and redundant .we use RS to reduce some related or redundant factors which can reduce the input nodes of networks and improve the interfere resistance. Under the premise of not affecting the training accuracy, it simplified the training of network. We use Rough Sets to reduce the collected index, and then use the reduced index as the nodes of WNN. The result proves that we can predict the cost quickly and accurately by using the model of RS-WNN.