MATLAB: Removing Duplicate Elements from Array sets

duplicate entrygenetic algorithmMATLABmatrix

I am constructing a genetic algorithm for finding optimal solution of a problem. The population used consists of three controller parameters (Kp Ki Kd)and fitness value. If the values of [Kp Ki Kd] are repeated the chances of finding a wrong solution due to premature conversion is high. So i want to check if there are any duplicate entries for [Kp Ki Kd] and remove them.

NOTE: I only want to remove if all the three [Kp Ki Kd] are same.

Best Answer

Let your array: K = [Kp Ki Kd]

Knew = K(any(diff(K,1,2),2),:);

Related Solutions

MATLAB: Find optimized solution of linear system of equations

Try this:

syms kd kp ki 
Eq1 =  0.1*kd +0.49 == 18;
Eq2 =  0.04*kd +0.1*kp +21.6 == 121;
Eq3 =   0.04*kd +0.1*ki +0.98 == 438;
Eq4 =   1.96*(kd +kp)+ 0.04*ki == 1030;
Eq5 =   1.96*ki == 1200; 
[A,b] = equationsToMatrix(Eq1,Eq2,Eq3,Eq4,Eq5);
Coeffs = lsqr(double(A), double(b));
kd = Coeffs(1)
ki = Coeffs(2)
kp = Coeffs(3)

producing:

kd =
           912.95079751536
ki =
          620.999261482952
kp =
         -397.442602821135

The Symbolic Math Toolbox sorts the variables in lexicographical order, so kd>ki>kp, thus the ordering in the results.

Experiment to get the result you want.

MATLAB: Unable to find explicit solution to differential eq

Hi Ditlev,

I believe this does have an analytic solution that is not overly complicated, of the form

x = (a+be^(-ct)) / (f+ge^(-ct))

First of all the scaling of this problem is not too good so I went to the variables (I used y in place of your x2)

u = km*t         x = (kp/km) x_old        y = (kp/km) y_old

in which case the equations are

x' = 2y - 2x^2
y' = -y + x^2

where differentiation is d/du. The starting value for the new x is

(kp/km)*1e-9 = 740.7

The most complicated the algebra gets is the roots of a quadratic eqn. The code below plots both a numerical and analytic solution.

Eventually the solution gets to the fixed point

x = r1 = 18.9966      y = -r1/2 + C = 360.8721

at which point x^2 = y and both derivatives are zero.

kp = 2*10^7;
km = 0.27*10^-4;
x0 = (kp/km)*1e-9   % starting point for x
y0 = 0;
% numerical version
[unum xynum] = ode45(@(u,xy) fun(u,xy), [0 .01], [x0 y0]);         % xy = [x; y]
%analytic version
C = x0/2;
r = roots([1,1/2,-C]);
r1 = max(r);
r2 = min(r);
alpha = 2*(r1-r2);
D = (x0-r1)/(x0-r2);
u = 0:1e-4:.01;
x = (r1-D*r2*exp(-alpha*u))./(1-D*exp(-alpha*u));   % solution
y = -x/2 + C;
figure(1)
plot(unum,xynum,u,x,'o',u,y,'o')
grid on
function dxy = fun(u,xy)
  dxy = zeros(2,1);
  dxy(1) = 2*xy(2)-2*xy(1)^2;
  dxy(2) = -xy(2) + xy(1)^2;
end

Best Answer

Related Solutions

MATLAB: Find optimized solution of linear system of equations

MATLAB: Unable to find explicit solution to differential eq

Related Question