WNN Nonlinear Modeling Method based on Heuristic RS Attribute Reduction Algorithm and Its Application
Download as PDF
DOI: 10.25236/icsemc.2017.35
Author(s)
Gan Xusheng, Qu Hong, Han Jun
Corresponding Author
Gan Xusheng
Abstract
Compared with the traditional neural network, Wavelet Neural Network (WNN) has many advantages, but it cannot achieve the expected modeling effect when dealing with the nonlinear modeling problem with large input dimension. In order to solve this problem, a WNN modeling method based on Rough Set (RS) attribute reduction is proposed. In this method, a heuristic attribute reduction RS algorithm is used to reduce the input variables in advance, and then the WNN model is established on this basis. The experiment result on the aerodynamic modeling of the aircraft shows that the proposed method is effective and feasible, so as to provide a good choice for the nonlinear modeling of complex systems.
Keywords
Wavelet Neural Network, Rough Set, Attribute Reduction, Nonlinear Modeling.