neurofuz.m

FISMAT accommodates different arithmetic operators, fuzzification and defuzzification algorithm, imp

function  [sys, x0] = neurofuz(t,x,u,flag,ctr,Tsamp)

% NEUROFUZ An S-function for discrete controllers with a sample time.
% 
% As opposed to the s-functions FUZZY11, FUZZY21 NEUROFUZ does not pass previous
% values of inputs and outputs through the X parameter. Another increase in speed
% is possible by creating a MEX-file. (See Release Notes to SIMULINK)
%
% The number of inputs to the NEUROFUZ block can be a vector of any size. This
% is possible with the SIMULINK version 1.3.
%
% This s-function is nearly identical to the demo template SFUNDSC2.M in the 
% /TOOLBOX/SIMULINK/BLOCKS directory of SIMULINK 1.3
% Only two parameters are added in the function head (ctr,Tsamp). 
% CTR = name of the controller m-file
% Tsamp = sampling time
% CTR and Tsamp are passed as function parameters in the S-function block menu
% as function parameters. See examples with FUZZY11 and FUZZ21 for more infor-
% mation.
% A simulink "testfile" for the s-function blocks is ctest1_s.m the corresponding
% controller file is ct1_c1.m.
%
% FSTB - Fuzzy Systems Toolbox for MATLAB
% Copyright (c) 1993-1996 by Olaf Wolkenhauer
% Control Systems Centre at UMIST
% Manchester M60 1QD, UK
%
%       23-November-1994
	
if abs(flag) == 2,
  % The FLAG==2 branch is at the top of the file for efficiency, since
  % the S-function will be called with FLAG==2 very frequently.
  % If FLAG==2, then SIMULINK is looking for the next point in real time. 

  if abs(round(t/Tsamp)-(t/Tsamp)) < 1e+8*eps,
    sys = feval(ctr,u,x,Tsamp);
  else
    % (SYS =) XNEW = X
    sys = x;
  end;

elseif flag == 3,
  % If FLAG==3, then SIMULINK wants to know what the next output is.
  % **** In this template system, y gets the current state x ****
  % (SYS =) Y = X
  sys = x;

elseif flag == 4,
  % If FLAG==4, then SIMULINK wants to know what TNEXT is.
  % TNEXT is the time when the integrator should be constrained to
  % call the system with FLAG==2.
  % First compute the number of samples so far
  numSamp = floor(t/Tsamp + 1e+8*eps);
  % Now sys returns TNEXT, the next time the discrete state should
  % be updated.
  % (SYS =) TNEXT = ...
  sys = (numSamp + 1)*Tsamp;
  
elseif flag  == 0,
  % This part takes care of all initialization; it is used only once.
  % The sizes vector is six elements long, and it looks like this:
  % sizes(1) = number of continuous states
  sizes(1) = 0;
  % sizes(2) = number of discrete states
  sizes(2) = 1;
  % sizes(3) = number of system outputs (length of output y)
  sizes(3) = 1;
  % sizes(4) = number of system inputs (length of input u)
  sizes(4) = -1;
  % sizes(5) = number of discontinuous roots; unused feature, set to zero
  sizes(5) = 0;
  % sizes(6) = direct feedthrough flag; used to detect algebraic loops.
  % Set sizes(6) to 1 if the output y depends directly on the input u.
  % Otherwise, it should be set to 0.
  sizes(6) = 0;
  % Set the initial conditions on the states
  x0 = 0;
  % (SYS =) SIZES = [0 1 1 1 0 0]';
  sys = sizes';

else 
  % Flags not considered here are treated as unimportant.
  % Notice that since there are no continuous states in this system,
  % there is no need to deal with FLAG==1.
  % Output is set to [].
  sys = [];

end     % if abs(flag) == ...