Data-driven techniques for direct adaptive control: the lazy and the fuzzy approaches

This paper presents an approach to modeling and controlling discrete-time non-linear dynamical system on the basis of a finite amount of input/output observations. The controller consists of a multiple-step-ahead direct adaptive controller which, at each time step, first performs a forward simulation of the closed-loop system and then makes an adaptation of the parameters of the controller. This procedure requires a sufficiently accurate model of the process in order to meet the control requirements. Takagi-Sugeno fuzzy systems and Lazy Learning are two approaches which have been proposed in control literature as effective ways of identifying a plant. This paper compares these two approaches in two main configurations: (i) when the number of observations is fixed and (ii) when new observations are collected on-line after each control action. Simulation examples of the control of the manifold pressure of a car engine are given.

[1]  Roderick Murray-Smith,et al.  A local model network approach to nonlinear modelling , 1994 .

[2]  M. J. D. Powell,et al.  Radial basis functions for multivariable interpolation: a review , 1987 .

[3]  W. Cleveland,et al.  Smoothing by Local Regression: Principles and Methods , 1996 .

[4]  Giorgio Picci,et al.  Identification, Adaptation, Learning , 1996 .

[5]  John Moody,et al.  Fast Learning in Networks of Locally-Tuned Processing Units , 1989, Neural Computation.

[6]  Karl Johan Åström,et al.  Computer-Controlled Systems: Theory and Design , 1984 .

[7]  Michio Sugeno,et al.  Fuzzy identification of systems and its applications to modeling and control , 1985, IEEE Transactions on Systems, Man, and Cybernetics.

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

[9]  Mauro Birattari,et al.  The local paradigm for modeling and control: from neuro-fuzzy to lazy learning , 2001, Fuzzy Sets Syst..

[10]  George Cybenko,et al.  Just-in-Time Learning and Estimation , 1996 .

[11]  Mauro Birattari,et al.  Lazy Learning Meets the Recursive Least Squares Algorithm , 1998, NIPS.

[12]  Hugues Bersini,et al.  Direct adaptive fuzzy control for MIMO processes , 2000, Ninth IEEE International Conference on Fuzzy Systems. FUZZ- IEEE 2000 (Cat. No.00CH37063).

[13]  Marco Saerens,et al.  Fuzzy adaptive control of a certain class of SISO discrete-time processes , 1997, Fuzzy Sets Syst..

[14]  R. H. Myers.  Classical and modern regression with applications , 1986 .

[15]  H. Bersini,et al.  Using incremental learning algorithms in the search for minimal and effective fuzzy models , 1997, Proceedings of 6th International Fuzzy Systems Conference.

[16]  Richard F. Gunst,et al.  Applied Regression Analysis , 1999, Technometrics.

[17]  Hugues Bersini,et al.  NLMIMO, Non-Linear Multi-Input Multi-Output Toolbox , 2000 .

[18]  Geoffrey E. Hinton,et al.  Learning representations by back-propagation errors, nature , 1986 .

[19]  David W. Clarke,et al.  Generalized predictive control - Part I. The basic algorithm , 1987, Autom..

[20]  Mauro Birattari,et al.  The Lazy Learning Toolbox. For Use with Matlab , 1999 .

[21]  Gianluca Bontempi,et al.  Local learning techniques for modeling, prediction and control , 2000 .

[22]  Roderick Murray-Smith,et al.  Multiple Model Approaches to Modelling and Control , 1997 .

[23]  Karl Johan Åström,et al.  Theory and applications of adaptive control - A survey , 1983, Autom..

[24]  J. W. Modestino,et al.  Adaptive Control , 1998 .

[25]  Y. D. Landau,et al.  Adaptive control: The model reference approach , 1979, IEEE Transactions on Systems, Man, and Cybernetics.