07-11-2011 09:46 AM
Hi All,
I am trying to do system identification of Solar water heating system. I have one input, two measured disturbances (which I have taken as inputs) and one output. I am simulating this system in another software. In that software, I send different random signals as inputs to get an idea of what the output could be. I used these results to perform system idebntification. The VI is attached. Now the problem is my VI does not give me close results of measured and simulated values of the output (when I perform system model verification). the both measured and simulated results should be almost overlapping but I am getting a huge difference at some points, which is not good. The waveform graph 3 shows these two value.
Details of the files attached:
Data File: The valve at index 0(column 1 is the input, values at index 1 and 2 are the measured disturbances and the values at index 3 are the output).
VI: I have used split VI to split my data i.e half for getting model and half for verification of that model.
Your help will be really appreciated.
Thanks in Advance......
Kindest Regards
Muhammad
07-14-2011 04:07 AM
Hello Muhammad,
Since we have this open as a Service Request, I thought I should update the forums with where we are.
I looked at what you are trying to do carefully, and I believe that it is the way you are using the signals that is the issue here, as otherwise your code is fine.
If you want to use system identification to extract a model for your system, you should only use pure signals for stimulus and ignore the disturbances. Your control system should eventually be able to handle disturbances, but you will not get a good model if you include them in this part of the process. You should only use the disturbances during the verification part.
Additionally, you are using your disturbences as stimulus signals themselves, which might not be a correct approach. The disturbances should be used in combination to your stimulus signals and not on their own, and again, they should only be considered during verification.
I hope this clarifies how the System Identification and Verification tools are best used.
Kind Regards,
Michael S.
Applications Engineer
NI UK & Ireland
Let me know if I can help further.