We are receiving 3 phase current readings from Induction motor using CT sensors. I am not aware of what processing is being done on the spot. My job is to do Motor Current Signature Analysis on it using python. But the currents have weird spikes everywhere. Sampling frequency is 6500, Fundamental Frequency = 50Hz

Plot of 1000 records from phase 1.

Phase 1 current

The Amp vs Freq plot

The Amp vs Freq plot

Tried PSD using welch transform:

enter image description here

Thought maybe welch transformation shows the peak at 50Hz so its working but then tried adding a frequency of 500Hz to it and it didn't show up on the plot.

Also tried using wavelet transformation to remove noise but the results are the same. At this point I am not sure how to proceed. Any suggestions are welcome. Thank you.

  • 1
    $\begingroup$ You have massive DC offset in the current that masks everything. Is that expected (DC motor) or a measurement error? It would also help if your label your axes: Is Y_axis in actual Ampere? Is the time axis in actual time or in samples ? $\endgroup$
    – Hilmar
    Mar 27, 2023 at 20:27
  • $\begingroup$ @Hilmar DC offset, it seems so. I substracted mean from all elements and the normalized the data (for convenience) and it seems that its now visualizing as expected. Thank you. X-axis represenst the samples. Y is Ampere. $\endgroup$ Mar 27, 2023 at 20:49
  • $\begingroup$ Also I didn't abs() the FFT values $\endgroup$ Mar 27, 2023 at 20:59

1 Answer 1


Credits to @Hilmar

It seems that the data had DC offset. To counter this, I subtracted mean from all elements and it seems to work as expected. enter image description here


enter image description here

So, Also, now I used absolute values of FFT.


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.