Design Method and Implementation of RBF Network Based on OLS Algorithm

In RBF network learning,using random data points as radial basis function centers to compute neural network weights,the result is often large network or unsatisfactory,and thereby cannot meet the requirements. Based on orthogonal least squares method,selecting radial basis function centers can find the centers one by one to build good network,and the maximum of the center selected by the algorithm can avoid ill-conditioned problem. The orthogonal least squares strategy is simple and effective,and the designed radial basis function network has a very good effect.