Artificial Intelligence (AI) Mastering Development

Feature extraction for exponentially damped signals

I am looking into exponentially damped signals where it is a stationary signal (after implementing the Adfuller statistical test) and I would like to look into how can I extract meaningful features out of the signal in order to do pattern recognition with machine learning. Can anyone guide me on where I can find articles/blogs of signal processing techniques and feature extraction of exponentially damped signal?

My situation:

I want to look into features that relate to damping of the signal, I already looked at it in the frequency domain and I found out that from my datasets (considering the first 3 Natural frequencies/modes) the peaks are almost the same (there’s like the same deviation by only like [+] or [-] 0.5 from freq. values). Looking into the damping factor, I found out that only the second damping ratio was different but still small deviation around the same ([+] or [-] 0.5). So, I thought that it would be difficult for machine learning to identify the difference between cases. One of my ideas is to look into energy dissipation as it might be related to damping, but I don’t know how to approach it or from which domain I need to go in order to get the features.

Side Question:

I have several questions regarding signal processing:

  • Say I have a signal and would like to extract features from it,
    what steps or points that I should know in order to implement signal processing? (As I am using Python).
  • When I used signaltonoise function online (python) in order to see
    the signal-to-noise ratio and I got a positive SNR. However, if I
    pass the signal into, for example, a band-pass filter to concentrate on a certain frequency band I get a negative SNR. Why is that?
  • How can I extract features from STFT? And I also know about wavelet and HHT, what are the uses of both algorithms and how to also extract features from it?

Leave a Reply

Your email address will not be published. Required fields are marked *