Page 1000 - The Mechatronics Handbook
P. 1000

0066_frame_Ch33.fm  Page 24  Wednesday, January 9, 2002  8:00 PM

















































                       FIGURE 33.34  (a) The surface obtained after first iteration, (b) Membership functions after learning for the variable
                       x 1 , (c) Membership functions after learning for the variable x 2 , (d) The surface after learning.






















                       FIGURE 33.35  (a) The position control with neuro-fuzzy controller (U  = 8 V), (b) The position control with
                       neuro-fuzzy controller (U = 10 V).

                         The adapting parameters (weights, centers, and spread for Gaussian function) must be persistent.
                       Declaring global, static or using the workspace in order to store the are useful techniques to accomplish
                       the task.
                         The newest version of Simulink offers the possibility to write wrapper S-function, to use the callbacks
                       functions, and as an alternative ADA or Fortran programming language.


                      ©2002 CRC Press LLC
   995   996   997   998   999   1000   1001   1002   1003   1004   1005