Power Spectrum density plot

i want to analyze my signal, and filter the noise from this signal based on the power specturm density plot. However, how can I know which filter and what frequecnies should i consider? Facing difficulty in interpretation of plot

of PSD plot

[Mario Malic]

There is a signal processing onramp course you can take and it can give you a bit insight on the topic itself. There are lots of things to consider and it all depends on the system you measured yourself.

pwelch does windowing, overlapping and averaging itself (if you do it with default values) and this looks already like a decent result itself that you can use to get the information you need.

You should have some values that you have calculated yourself so you can look for them in the graph. For example, these 4 peaks from 3 - 5 * 10e7 Hz seem like they are harmonics of some natural frequency of the system. Maybe, due to pwelch settings, you can't see them in the frequencies 1.3 - 3 * 10e7 Hz.

These peaks at 3.1 and 6.2 * 10e7 Hz, could also be harmonics of the 0.2 * 10e^7 Hz.

It's you who has to interpret the results and match them to your findings.

Maybe someone else can chip in with more knowledge, I would like to hear more opinions.

[Mathieu NOE]

would be helfull to know what you measure and what you want in terms of information

is it the mean peak at 3 MHz ? you want to bandpass it to remove below and above that frequency ?

[MANINDER CHOUDHARY]

This is my signal, I want to remove noise from this signal. So I plotted my signal as Welch's power spectral density plot, to see where the signal has high power, based on that I can extract the origninal signal by removing the noise. I'm confused how i design or implement filter that give me only denoised signal.

Answer 1

[Pooja Kumari]

Dear Maninder,

It is my understanding that you are facing issues with analysing the signal and choosing the filter for removing the noise from the signal based on power spectrum density plot.

You can look for frequency components with higher power values as the peak or spikes in the Power spectrum density plot. These peaks indicate the presence of significant frequency components in the signal. By examining the height and location of these peaks, you can identify the frequencies that contribute the most power to the signal. These are the frequency components of interest.

The PSD plot can reveal distinct frequency bands or regions of interest where the power is concentrated. From the graph plotted, you can see that the major signal of interest lies between 0 – 1.5e7 Hz frequency.

Hence, you can apply a low-pass filter to cutoff the frequency below 1.5*10^7 Hz, to filter noise from the signal. You can refer to the documentation below for more information on low-pass filter:

https://www.mathworks.com/help/signal/ref/lowpass.html

I hope this helps!

Regards,

Pooja Kumari

[MANINDER CHOUDHARY]

Thank you Pooja for the detailed explanation.