1
$\begingroup$

I am working on a sun-synchronous satellite Simulink model wich determines its attitude from a Sun sensor and a star sensor whom measure the sun and star direction in the satellite body reference frame. I have to introduce measure errors on both sensors using a "white noise" Simulink block, but I prefer to generate the noise through a Matlab function block using the function "rand" (or shoul'd I use the "randn" function?).

My issue is that I have to filter the noises on the measurements before sending them to a block who compute the attitude quaternion error to feed it to a PID controller blockset which commands a Reaction Wheels Assembly. Without noise the satellite flies smoothly.

I have tried some solution such as lowpass filter, Butterworth filter, mean on peaks and downs, but nothing seems to work properly.

I cannot figure out, for example, which should be the cut off frequency to set up on a lowpass filter ( I cannot figure out if the lowpass filter is the correct choice for such a problem, too).

Improve this question
$\endgroup$
1
  • $\begingroup$ You should use randn for gaussian n, not randn Second of all, you do realize your whole PID control loop is a low-pass filter, right? $\endgroup$
    – Ben
    Oct 29, 2019 at 16:06

1 Answer 1

Reset to default
1
$\begingroup$

It may be useful to have a deeper understanding of you satellite's dynamics. What I mean is that by observing your attitude quaternion error variation DSP ( or using the motion mathematical model), you may be able to choose a convenient filtering type/tuning to get rid of most possible white noise.

Improve this answer
$\endgroup$
4
  • $\begingroup$ I could place a lowpass fiilter befor the derivative control of the PID, but I am not able to select a cut-off frequency. Introducing the nise in the sensors measures "make to explode" the simulation rsulting in an uncontrollabe satellite. I would like to know how to get rid of the noise in the measures or how to render unsensitive the PID control to the noise. $\endgroup$
    – user45909
    Oct 29, 2019 at 14:01
  • $\begingroup$ I understand your concerns, My point was that tuning is different from an extreme case where the satellite is rapidly spinning around itself or a more smooth satellite motion case. In addition, the white noise will be covering all your useful signal's frequency band. Thus, you may not be to get rid of it with a lowpass filter. Have you considered averaging filtering ? This will help reduce the noise level and may solve the problem. Anyway, you may want to check this source about filtering options in a control system controlguru.com/using-signal-filters-in-our-pid-loop $\endgroup$
    – M.FAKHFAKH
    Oct 29, 2019 at 14:20
  • $\begingroup$ Here is a reseach gate discussion about how to white noise filtering methods depending on the tradeoff: researchgate.net/post/… . $\endgroup$
    – M.FAKHFAKH
    Oct 29, 2019 at 14:35
  • $\begingroup$ Thank you.My satellite is spinning on the pitch axis once per orbit, so it is a slow spinning satellite on one axis , and no spinning on the other axes. I will check your proposal. Thank you so much! Em $\endgroup$
    – user45909
    Oct 29, 2019 at 14:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.