Minimum input data resolution

Dear all,

I'm trying to use the system identification toolbox for the first time for characterizing a microgrid. While preparing the input data for the model (in this case I have 1 second time resolution data), I asked myself whether or not this time resolution I have is sufficient or not. Can any of you provide me with some insight as to how I could try to asnwer this question?

Of course, it is not practical for me to get data with higher time resolution than 1 second. But even in this case, I would like to have a feeling to what kind of inaccuracies that I might get for not having a higher time resolution. How can I get an estimation on this before starting the modeling process?

regards

Buddi

Answer 1

[Rajiv Singh]

Look up Nyquist Sampling Theorem. If you are sampling (hopefully with anti-aliasing) at 1Hz then you cannot theoretically capture dynamics higher than 0.5 Hz. Some (may be subtle) aspects:

• This is in linear case only.
• In practice it is good to keep sampling 5-10 times faster than the highest frequency in the system that you are interested in capturing.
• This is in absence of any additional priors; otherwise: look up compressed sensing literature.

[Charitha Buddhika Heendeniya]

Thanks Rajiv