Position error of motor shaft

I want to obtain the position error of motor shaft so I subtracted the estimated position from the measured position. modulus function is used so the estimated and the measured positions vary between 0 to 2pi (electrical radians) as shown in the figures attached. the problem is that the error becomes 2pi at instant times. What can I do to not show this big error (2pi) in the figure?

Answer 1

[Mathieu NOE]

hello

you can make 2 pi complements on the error signal , following that logic

if error < -pi then error = error + 2pi

if error > pi then error = error - 2pi

goal is to keep error witin -pi / + pi bounds

[saleh shlimet]

Thank you for your answer as I conducted all my experiments and saved all data (results) in the workspace as shown in the attached. All position error data is saved as in the figure I sent to you. How can I use the data saved in the workspace to implement your suggestion? In other words, can I use the command window to write a code or algorithm to redraw all position error data by using your idea?

[Mathieu NOE]

hello

I was meaning that this logic must be implemented directly in your simulink model

where I draw the red rectangle

[saleh shlimet]

Hi Mathieu, Thank you for reply. I now that but to do that I need to conduct the experiment again for all results and this will cost me a lot of time. I saved all my results (position errors) as workspace. So the position_error where you draw the red rectangle is saved in the workspace as I showed you in the figure. My question is how can I implement the logic to this saved position error data in the command window to redraw the position error?

[Mathieu NOE]

ok I understand

my remark was triggered by the fact that you may use error psoition signal in a feedback loop and of course those angle jumps will affect the entire system behaviour

if this error position signal is simply for visualization,why not simply make a matlab script and run it on your already available data ?

[saleh shlimet]

What do you mean by matlab script and how can I do it?

[Mathieu NOE]

can you share the data ? (as mat file)

[saleh shlimet]

I got it but I had an error when I tried to plot the position error (Operator '>' is not supported for operands of type 'timeseries') as shown in the attached

[Mathieu NOE]

convert your timeseries data to a standard numerical array

[Mathieu NOE]

If M is your time series

V=M.Data

[saleh shlimet]

I tried it but it gave me the position error without removing 2pi see attached it seems that the code in the script was not excuted. I sent you the script file and error position data.

[Mathieu NOE]

yes , I understand

the way you wrote your function is ok if you do it for one sample at a time (like you would implement if in a simulink block)

as we deal here with a long array (record) you have to it differently :

load('matlab_error.mat')

t = pos_error_reg.time;
err = pos_error_reg.Data;
err2 = pos_error(err);
plot(t,err,'b',t,err2,'r')

%%%%%%%%%%%%%%%%%%%%%
function r = pos_error(a)

    r = a; % replaces "else" case 
    
%   if a > pi  % condition 1
    id1 = a > pi;
    r(id1) = r(id1) -2*pi;
    
%   elseif a < -pi  % condition 2
    id2 = a < -pi;
    r(id2) = r(id2) +2*pi;


end

[saleh shlimet]

Thank you very much Mathieu it is working properly.

[Mathieu NOE]

My pleasure !