Non-linear Adaptive Predictive Control based on orthogonal Wavelet Networks

摘要

In this paper, a nonlinear adaptive predictive control strategy based orthogonal wavelet network model only using scaling function is presented. Based on a set of orthogonal wavelet functions, wavelet neural network performs a nonlinear mapping from the network-input space to the wavelons output space in hidden layer firstly. Then, the output layer adopts linear structure. Its weight coefficients, i.e., can be estimated by a linear least-squares estimation algorithm. The excellent statistic properties of the weight parameters of wavelet network as linear least-squares estimation algorithm in system identification can be obtained. Based on developed recursive algorithm, a single input - single output nonlinear adaptive predictive control strategy is implemented. A simulated CSTR process example is used to illustrates the application of the control scheme.