Optimization Study Of The Hidden Structure And Parameters In The RBF Neural Networks

Among Artificial Neutral Network methods, Radial Basis Function Neural Network (RBF-NN) is the one of the most popular method. The difficulty in applying RBF-NN lies in building an appropriate optimal construct, optimal number of hidden neurons, and optimal parameters (centers, widths, and weights).A modified RBF-NN is proposed in this text, which is integrated with the K-Means clustering based on the Rough sets theory (Rough K-Means), and EM-Clustering combine with Multiple Linear Regression (MLR) algorithm respectively, and two methods are applied to eliminate redundant information of RBF network and improve the predicting performance through optimizing the structure and the weights and bias. And developing naphtha dry point soft sensor is employed to illustrate the performances of three modified RBF-NN. The results show that the predicting performance of the soft sensor is improved and then decreased with deleting the redundant nodes. Comparing with the other methods, EM&MLR is more accurate and simpler.