Adaptive neural network structures for non-linear process estimation and control

Abstract While modern feedback controllers are based on linear models, many processes of concern to chemical engineers are non-linear. The models for these processes are often too complex for non-linear controller design, and their linearized counterparts may not adequately represent the process dynamics. Adaptive radial basis function neural networks have been implemented in process estimators and controllers to approximate non-linear functions that describe the process. Existing algorithms have been proposed which require specification of network properties such as dilation of the basis functions. This work demonstrates that the function representation may be improved by selection of an optimal dilation, and presents an algorithm that allows simultaneous adaptation of dilation and node weights. In order to deal with functions that may have more than one dominant dilation, a multiresolution network adaptation algorithm is proposed. Lyapunov stability is proven for both strategies, and performance is evaluated for the control of an exothermic CSTR.

[1]  Jun Zhang,et al.  Wavelet neural networks for function learning , 1995, IEEE Trans. Signal Process..

[2]  Evanghelos Zafiriou,et al.  Robust process control , 1987 .

[3]  Ali Cinar,et al.  Approximate Dynamic Models for Chemical Processes: A Comparative Study of Neural Networks and Nonlinear Time Series Modeling Techniques , 1992 .

[4]  Gilbert Strang,et al.  Wavelets and Dilation Equations: A Brief Introduction , 1989, SIAM Rev..

[5]  Bhavik R. Bakshi,et al.  Wave-Nets: novel learning techniques, and the induction of physically interpretable models , 1994, Defense, Security, and Sensing.

[6]  Robert M. Sanner,et al.  Gaussian Networks for Direct Adaptive Control , 1991, 1991 American Control Conference.

[7]  Geoffrey E. Hinton,et al.  Learning representations by back-propagating errors , 1986, Nature.

[8]  Bhavik R. Bakshi,et al.  Wave‐net: a multiresolution, hierarchical neural network with localized learning , 1993 .

[9]  George Cybenko,et al.  Approximation by superpositions of a sigmoidal function , 1992, Math. Control. Signals Syst..

[10]  Robert M. Sanner,et al.  Enhanced Algorithms for Adaptive Control Using Radial Gaussian Networks , 1992, 1992 American Control Conference.

[11]  Dale E. Seborg,et al.  Experience with nonlinear control and identification strategies , 1994 .

[12]  F. Girosi,et al.  Networks for approximation and learning , 1990, Proc. IEEE.

[13]  Jean-Jacques E. Slotine,et al.  Space-frequency localized basis function networks for nonlinear system estimation and control , 1995, Neurocomputing.

[14]  N. V. Bhat,et al.  Use of neural nets for dynamic modeling and control of chemical process systems , 1990 .

[15]  Hector Budman,et al.  Developmental studies of an adaptive on-line softsensor for biological wastewater treatments , 1999 .

[16]  Manfred Morari,et al.  Robust Controller Design for a Nonlinear CSTR , 1989, 1989 American Control Conference.

[17]  Dale E. Seborg,et al.  Nonlinear internal model control strategy for neural network models , 1992 .