08-13-2007 03:27 AM
08-13-2007 07:59 AM
if you use different IO modules you should synchronize them, so that all data has the same time axis. synchronization has to be made for the start of the IO (use the same trigger signal for all cards) and the clock signal. otherwise you compare data from different time axis which may lead to wrong results.
what IO modules do you use?
take a look at the examples ..\examples\DAQmx\Synchronization shipped with LV.
08-13-2007 09:15 AM
RRapha,
Generally speaking from the System Identification point of view (the algorithm), you should always measure each output of your system at the same sampling time, all synchronized with the stimulus signal that you applied to the inputs. If you try to just measure one input-opt per time, you can have interactions between other channels that are not captured in the experiment.
Notice also that the type of excitation applied to your system is very important to identify the model when you try to identify MIMO models. If you try to apply the same signal to all inputs, you can have a modes of interaction that are excited at the same time and you can't distinguish between the experiment. The type of signal is highly dependent on the type of system that you are trying to identify. This paper below discuss different signal and gives you an idea to what signals can be used for the application below. For other application, probably other paper can better describe the signals that should be applied:
http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20070018341_2007018863.pdf
For SIMO systems, since you have only one input, if you use the MIMO model and add the missing dimension of the input. This should work for your application.
Hope this helps.
08-13-2007 09:40 AM
Hi Barp,
first off all thx. i would like to specify my question. Due to the coupling in the test stand ( rotor in active magnetic bearing)
we can reduce the system to 2 DOF -> a 2x2 MIMO system. But as we´re not able to excite the system with to signals at the same time we only can measure
SISO or SIMO. I tried the MIMO-Estimate SS-Model with 2 stimulus and 2 responses. As I´m only able to excite the system at one of these point
i added the second stimulus as an array full of zeros in order two get a SIMO model. But the result was bad.
To get a p-canon 2x2 Modell of my Test stand is there theoretically a difference in the result If i estimate 4 times TF SISO after another or estimating SIMO 2 times?
Thx and greetings,
RRapha