Compose signal processing - PSD 2 Time-serie

cimes

Hi all,

I am trying to transform a PSD into a time-history. I am using the method described in the E-book of signal processing under Compose but I am facing two errors in the script, namely I cannot display the time signal and the inverse calculation to recalculate the DSP with the time signal does not work with pwelch command

DSP2Time.oml

.
I suppose that this is due to the fact that the vector of the time signal contains ascii characters (NaN + NaNi) but I cannot find the solution.
I am attaching the script.
Thank you in advance for your help.
Arnaud

RSGarciarivas

Hi Arnaud. Some time ago I worked on a very similar tool. Let me look for it and I'll have a look and share it.

Rafael

RSGarciarivas

Hi Arnaud. Have a look here:

PSD to Time Series OML Utility — Altair Community

I recovered this old utility. If you look at the callback functions, you'll see how to take an input PSD and generate a time-domain realization.

Rafael

cimes

Hi Rafael,

Thanks again for your help.
I still have a little problem, my DSP is not in db but in acceleration ((m/s2)^2). So I deleted the db2mag and mag2db commands but I find the problem in the creation of the time signal.
I attach the Excel csv file of my DSP.

Time.zip

Thanks.

Best regards.

Arnaud

RSGarciarivas

Hello Arnaud,

Sorry for the delay in my response. I've had a look at your files and seems to me that your PSD is in the appropriate format to use it as is in the PSD2Time utility.

The PSD is displayed in dB in the plot for visualization purposes, but the input file should contain magnitude data, not dB.

Running your file without changing the script, it looks fine on my side:

Let me know if I'm missing anything.

Rafael

cimes

Hi Rafael,

I ran the script without changing anything and here is the result, I don't have exactly the same thing as you:
Between 0 and 2Hz we must have zero values, like after 60Hz
We must have linear sampling between 2 and 5Hz and between 5 and 60Hz.
I wonder if the amplitudes of the time signal are correct given that the initial PSD directly includes the amplitudes of the accelerations (in m/s^2)^2?

When you run the script twice, it is impossible to define the output directory again.

Thanks again for your help.
Have a good weekend.

Regards.

Arnaud

RSGarciarivas

Hi Arnaud. Thanks for bringing this to my attention. I've identified what I have to fix, when I'm resampling the PSD and the length of the desired signal is greater than 1, the limits of the PSD are not respected.

I'll try to fix it either today or on Monday.

Regarding the issue of the output location, it works fine on my side, can you double check, please?

cimes

Hi Rafael,

I have just restarted the test which consists of launching the script several times and indeed the selection of the output directory is done correctly.
I think that the problem occurs when there is a problem in the execution of the script.

Thanks.

Have a nice day.

Arnaud

RSGarciarivas

Hi Arnaud,

I've uploaded a new version of the utility here:

PSD to Time Series OML Utility — Altair Community

I think the issue has been fixed. Let me know if not.

Rafael