For loop iterations random noise

Hi everyone,

My problem has to do with for loops and random noise. What i am ultimately trying to do is add noise to a signal so i can find the phasors (but that's beyond the point) When i add noise in an equation i feel like the same noise coefficient is being used for the 100 iterations of the equation. Is there a way when i run a for loop that i get a different random coefficient? This is my code so far

%Synchrophasors Homework #4
%Chris Coronel
%Problem #1

clear all;
close all;

j = sqrt(-1); %imaginary variable

N =24;  %Number of samples per cycle

f1 = 60; %Nomincal frequency

f2 = 60.5; %Off nominal frequency

del_w = 2*pi*(f2-f1); %Difference between off and nominal frequency

w_prime = 2*pi; % The rate of change of the frequency

del_T = 1/(N*f1);  %Sampling interval at nominal frequency

phi_0 = 0.1; %Assumed initial position of phase angle

n = 0:95;

Phi_1 = (phi_0 + del_w*n*del_T + (1/2)*w_prime*(n*del_T).^2)';%Equation to sample at
%%%%%This produces the exact same phase angles as Table 4.1 w/o noise%%%%

Phi_2 = (phi_0 + del_w*n*del_T + (1/2)*w_prime*(n*del_T).^2 + normrnd(0,.01))'; %Equation to sample at
%%%%%This produces the sample of phase angles with noise distribution%%%%%

%%%%Create the Coefficient Matrix%%%%%%

for m=0:95; %Number of samples per cycle interval

B1 = 1.^m; 
B2 = m*del_T;
B3 =(m*del_T).^2;
format short

B(m+1,:) = [B1 B2 B3]; %Coefficient Matrix starting at m = 0

end

B_T = B'; % Transpose of Coefficient Matrix
%%%%Solve recursively for the phase angle%%%%%%

for p =1:(96-N);
    A1 = inv(B_T(:,p:N-1+p)*B(p:N-1+p,:))*B_T(:,p:N-1+p)*Phi_1(p:N-1+p,:);
    A2 = inv(B_T(:,p:N-1+p)*B(p:N-1+p,:))*B_T(:,p:N-1+p)*Phi_2(p:N-1+p,:);
    A_no_Guassian(p,:) = A1
    A_Guassian(p,:) = A2
end

delta_f1 = mean(A_no_Guassian(:,2)/(2*pi)); %difference of off and nominal freq mean no noise
f_prime1 = mean(A_no_Guassian(:,3)/pi); % mean of rate of change of frequency no noise

delta_f2 = mean(A_Guassian(:,2)/(2*pi))%difference of off and nominal freq mean with noise
f_prime2 = mean(A_Guassian(:,3)/pi) % mean of rate of change of frequency with noise

You will notice that the delta_f2 matches the delta _f1 which is not suppose to be the case. The delta+f2 should have some error since we added white guassian noise. Can someone help please, it'll be much appreciated !

Warm regards Chris Coronel

Answer 1

[Matt Tearle]

Phi_2 is Phi_1 plus a single random value. Without delving into the details of the ensuing math, I'm going to guess that's what's causing the problem. Try

Phi_2 = Phi_1 + normrnd(0,.01,size(Phi_1));

instead.

Also, pay attention to the Code Analyzer messages... Bad things are happening in that loop :)

[Christopher]

Thanks for the input Matt, much appreciated!