I would like to classify the signals from measurement. After seeing the results of the psd graphs, the graph area of each sample is signal is different.
I think I can classifiy the sample from the area of the signal in the freqwency domain (between 200-800Hz).
I have try to use intergration.vi but the result is not change significanly, eventhough there are any change in the graph.
Is there something wrong with my code or the value is to small so I cannot see the difference?
Another approach to classify is to summing the amplitudes of the frequency (from 200-800hz), but I still don't have any idea to find this approach.
It is really difficult to tell by looking at a small image of a diagram. It is even more difficult to tell without seeing the data.
Please run your VI so that some of your typical data is shown on the indicators. Stop the VI. Then Edit >> Make Current Values Default. Save. Post that VI so we cen see the data.
Thank you for the reply, and sorry for the delay response.
Here is my graphs, if we see the graph there is a different curve for each section in freq domain.
It is hard to tell by looking at the graphs. We do not have all the settings for the acquisition, filtering, and integration. Also some people will not open any files except LabVIEW files or text files with data for security reasons.
Please follow the directions I gave in the previous post to save the data in the VI as default and then post the VI. Then we can have actual data which can be manipulated.
I notice that the impulse in the first 8-10 ms seems to be very similar in all the graphs. You might get better results if you remove that part from the data before you start the analysis.
here I attach the my vi file, I have try to use sum of array to find the total amplitudes of the graph.
Sum is not very meaningful because some elements are positive and ohters are negative. RMS is probably a better measure.
The data set in the VI you posted has almost nothing significant in the 200-800 Hz band. The overall spectrum shows a rolloff with increasing frequency but no "interesting" spectral features. The raw data has an average DC component of about 5.7 with a peak to peak variation about one tenth of that. Removing the mean before other processing may be useful.
You may not need a filter at all. If you take the RMS value of the spectral components representing 200-800 Hz or whatever bandwidth you need, it is about the same as the RMS of the spectrum of the filtered data.
In the attached version I placed a constant obtained from the Waveform indicator inside the Enabled case of a Diagram Disable structure. The disabled case contains the DAQ Assistant. The mean is subtracted to get rid of the large DC component. The RMS of spectral components from 200 to 800 Hz is calculated.