1
$\begingroup$

i intend to extract features from time-domain measurement data. I feed the features to machine learning algorithms to detect anomalies.

In the time-domain, i extract mean, RMS, skew and standard deviation. I also want to execute a fourier transform and extract the features from the fourier transform. Intuitively, i would pick the mean frequency and the peak frequency for different frequency bands.

Unfortunately, i cant find any literature on the topic or other people who extracted features from fourier transform (and wavelet, cepstrum, Hilbert, ...) who are smarter than me. can anybody help?

$\endgroup$
1
  • $\begingroup$ It seems like you’re losing a ton of information by extracting only those features from the time domain. You’re not treating the data as a time series, and the features you plan to extract from the frequency domain are not such a remedy. It might be more helpful if you describe the machine learning problem you want to solve. // Time series are well-studied in machine learning. You might get some results by looking into speech recognition. $\endgroup$
    – Dave
    Nov 12, 2021 at 13:48

1 Answer 1

1
$\begingroup$

You can use the actual spectrum as your features. E.g. only select the lowest 10,20,30 frequencies. This approach has been used for e.g. this paper

$\endgroup$

Your Answer

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

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