# How do i do moving average in s-function level 2?

2 views (last 30 days)
dab483 on 18 Mar 2013
Hi i want to do moving average in s-function level 2, When using matlab script i wrote this formula where r=residual with size 1, samples is the time step.
window=10;
if samples > window
sumS = 0;
for l = 1:window,
if l==1,
S = (1/window) * (Htime(samples,:)) ./ R.^(1/2);
else
S= (1/window) * sum( (Htime(samples-(l-1):samples,:)) ./ R.^(1/2) );
end;
sumS=sumS + S*S';
end;
CovTime=((1/window)*sum(r(:,samples-window+1:samples)./ R.^(1/2))).^(2);
CovEnsemble=S*Pold*S' + (1/window);
CovDifference = CovTime - CovEnsemble;
if CovDifference > 0
Qparameter = CovDifference / sumS;
else
Qparameter = 0;
end;
the problem is when i am using s-function level 2, where i want to get this residual ~ r(:,samples-window+1:samples) , where i only have 1 data only when running the simulation. how do i get the previous data (residual) in s-function and to use in concurrently?
How do i define/ do it in s-function level 2? Thank you.

Kaustubha Govind on 18 Mar 2013
It looks like you need to maintain a "state" on the block. Please look into Using DWork Vectors in Level-2 MATLAB S-Function.
dab483 on 18 Mar 2013
Hi, i am a bit confuse, if in matlab script to run the program i can just put for example
for sample: 1:L (where L is simulation length) EKF algorithm window =10 if samples >window ...... end
so if i define L=200 , means that i have 200 samples but in s-function with the sampling time and everything, the length becomes larger. I can just put the sampling time=1 (which is big) but it wont work in my simulink model. any advice?