Application of RBF Neural Networks Based on a New Hybrid Optimization Algorithm in Flotation Process
An inferential estimation strategy of quality indexes of flotation process based on principal component analysis (PCA) and radial basis function neural network (RBFNN) is proposed. Firstly, the process prior knowledge and PCA method are used to simplify the networks’ input dimension and to choose the secondary variables. Then a new hybrid optimization algorithm of RBFNN is developed. The algorithm includes simplified rival penalized competitive learning method (SRPCL) to make an adaptive clustering of networks’ input pattern and recursive least squares method (LSM) with forgetting factor to update networks’ weights. The simulation results show that this inference estimation strategy has high predictive accuracy in flotation process.
KeywordsInput Pattern Radial Basis Function Neural Network Flotation Process Connective Weight Radial Basis Function Neural Network Model
Unable to display preview. Download preview PDF.
- 3.Wang, X.D., Shao, H.H.: The Theory of RBF Neural Network and Its Application in Control. Information and Control 26(4), 272–284 (1997)Google Scholar
- 4.Wang, X.F., Wan, Z.Q., Song, W.Z.: A New Hybrid Recursive Learning Algorithm for Ra-dial Basis Function Neural Networks. Control Theory and Application 15(2), 272–276 (1998)Google Scholar