New dynamic RBF neural network controller

It isn´t very effective to use RBF neural network as controller to deal with dynamic systems. So a new dynamic radial basis function network including feedback unit is proposed. The universal approximation theorem of DRBF is proved according to Stone-Weierstrass theorem. The intelligent controller based on this dynamic network is employed to deal with hydraulic position servo system. And learning algorithm based on integrative object function is deduced. The experiment results show that the intelligent controller has adaptability and robustness, and controller´s design does not depend on the system´s model.