Page 38 - Neural Network Modeling and Identification of Dynamical Systems
P. 38

26          1. THE MODELING PROBLEM FOR CONTROLLED MOTION OF NONLINEAR DYNAMICAL SYSTEMS




















                         FIGURE 1.3 A direct adaptive control scheme: r(t) is the reference signal; u(t) is the control; y(t) is the output of the
                         controlled object; y m (t) is the output of the reference model; θ c (t) are the adjustable parameters of the controller; ε(t) =
                         y(t) − y m (t) is the difference between the outputs of the object and the reference model (From Gang Tao. Adaptive control
                         design and analysis, Wiley-InterScience, 2003).

















                         FIGURE 1.4 Indirect adaptive control scheme: r(t) is the reference signal; u(t) is the control; y(t) is the output of the
                         controlled object; θ p (t) are the estimated parameters of the object;   θ p (t) is the estimate of object parameters; θ c (t) are the
                         adjustable parameters of the controller (From Gang Tao. Adaptive control design and analysis, Wiley-InterScience, 2003).


                         ture of such a system is shown in Fig. 1.3.In  rameters for the controller θ c (t). Estimates   θ p (t)
                         the direct adaptive control systems, the param-  are produced operatively (on-line), in the pro-
                         eters of the controller θ c (t) are adjusted by the  cess of object functioning, by computing the
                                                                                             ˙
                         algorithm implemented by the adaptation law,  value of the derivative   θ p (t) or the difference
                         which computes the values of the derivative    θ p (t + 1) −   θ p (t). The structure of such a system
                          ˙ (t) or the difference θ c (t + 1) − θ c (t).Thiscom-
                         θ c                                          is shown in Fig. 1.4.
                         putation is based directly on the tracking error  In both direct and indirect adaptive control
                         value ε(t) = y(t) − y m (t).                 schemes, the basic idea is that the ideal values
                            In the indirect adaptive control systems, the  of the controller parameters (for direct adaptive
                         parameters of the controller θ c (t) are computed  control) or the object (for indirect adaptive con-
                         using the coupling equation that maps an esti-  trol) are used as if they were parameters of a real
                         mate of object parameters   θ p (t) to values of pa-  controller or object, respectively. Because the ac-
   33   34   35   36   37   38   39   40   41   42   43