The Study Of Wavelet Neural Network And Their Application For Chemical Engineering Modeling

Accurate models are important to the research and. application of chemical engineering process. However, most problems in chemical engineering process are complex and we know little about their principles. So it is difficult to build accurate models directly by the principles. Neural network build models without the principles , it modeling chemical engineering process by sample data. The main focus of the thesis is on improving the learning algorithms of wavelet neural network (WNN), and optimizing the structure of WNN. The article includes the following parts mainly:(1) The originate and developments of neural network are reviewed, and several typical neural network are introduced, including MLFN(Multiplayer Feedfoward Network) Hopfiled Network Grossberg Network SOFM (Self-organized Feature Mapping) Network Recurrent Network RBFN(Radial Basic Function Network) Fuzzy Neural Network. The applications of neural network in chemical industry are introduced.(2) The originate and developments of wavelet and wavelet neural network are reviewed, and the properties and approximation ability of WNN is researched, the characteristics of consequent wavelet neural network and discrete wavelet neural network are introduced, and the parameters learning and training algorithms of WNN are improved. The applications of wavelet neural network in chemical industry are introduced.(3) According to the approximation ability of WNN, the nonlinear system is able to be modeled, obvious effect was shown in process modeling of residuum cracking and prediction of asphaltum yield and pyrolysis gas yield.In the end of this paper, we make a summary and describe the furture works.