[[alternative]]The Study of System Identification and Control Using Differentiable CMAC

[[abstract]]The Cerebellar Model Articulation Controller(CMAC) is a neural network model, which is viewed as a Basis Function Network(BFN). The conventional CMAC uses constant basis function is replaced by Gaussian Basis Function(a differentiable function) for retrieving the derivative information and to improve the generalization ability of CMAC. Not only the weights but also the means and variances are adjusted during the learning process. The new diferentiable CMAC network control system is established. The mathematical foundation for the new scheme is derived and the learning algorithm for control system is provided. After applying the new technique in some plant models, system identification and control performed to demonstrate the improvement of accuracy and the capability in providing derivative information.