Help with pressure wave analysis

1. Aug 10, 2010

swishidge

For my work I am analyzing pressure waves. I am new to this however, and do not know very much. I have attached two plots of the same wave. One plot is a line plot, and the other is a dot plot. For my analysis, I am using MATLAB.

As can be seen from the two plots, my data is very noisy. There is nothing I can do about this however, so I will have to work with what I am given. I would like to use a filter, a moving average, or perhaps a combination to better understand the data, but I do not know which to use, why one would be better than the other, advantages, disadvantages, etc.

Any help would be appreciated.

Thanks,

Nate

Attached Files:

File size:
26.8 KB
Views:
93
• dotplot.pdf
File size:
30.9 KB
Views:
94
2. Aug 10, 2010

minger

Without knowing your problem, it's hard to say which "peaks" are real and which ones are noise. Typically when I look at pressures in a CFD problem, I will perform FFT analysis afterward to see what frequencies show up in the solution.

Can you explain a little more about your setup and what exactly you're measuring?

3. Aug 10, 2010

minger

Without knowing your problem, it's hard to say which "peaks" are real and which ones are noise. Typically when I look at pressures in a CFD problem, I will perform FFT analysis afterward to see what frequencies show up in the solution.

Can you explain a little more about your setup and what exactly you're measuring?

4. Aug 10, 2010

swishidge

I am measuring the blast wave from a 105mm cannon. The data was collected using a static pressure gage. The pressure gage was located at the level of the cannon 10m away at a 45 degree angle, so the gage is actually 14.14m from the muzzle.

I tried performing an fft analysis and the attachment will show what I came up with.
This is the MATLAB code that I used:

figure
plot(fft(yvalues))

I don't think this is right though. I will look further into fft analysis.

Attached Files:

• fftattempt.pdf
File size:
42.5 KB
Views:
89
5. Aug 10, 2010

minger

No, most FFT plots I've seen/done are a simple line plot. On the x-axis you have frequency, and on the y-axis, you have magnitude. For example, if you're data plot was really y=sin(wt), on the FFT plot you would show a spike on the x-axis at w, and ideally zero everywhere else.

Unfortunately, you only have one "blast" from the cannon. You need at least a few pulses to get an accurate FFT.

What is the goal here? Are you simply trying to remove the noise?