08-17-2011 07:50 AM
the attached VI is the Model predictive control that I am trying to develop. I have three inputs (one manipulated and two measured disturbances). I have got my model from system identification. The model is discrete. I am just trying to tune it but could not do it properly. I am just wondering is there any problem with the model (as the D matrix is not zero, I have to make it zero for MPC or not? )or there is anything else wrong?. The limits on my inputs are; for input 1 it is from 0-210, for input 2 it is 10-60 and for input 3 it is 0-1500. The second and third inputs are measured disturbances so we can’t manipulate them in order to get the required output (I think we can set up this in weight matrix). The output has range of 20-60. I want my output to follow the refrence output for example 40, but it does not happpen....
@barp. Andy C. please have a look.
Looking forward to hearing from you.
08-18-2011 05:07 AM
Thanks for using the forums!
I have had a look at your code and it seems that you have set this up correctly. Which would suggest that there could be an error in your model.
I can't claim to be an expert in MPC design although there is a pretty comprehensive application in the Example Finder (Help » Find Examples...); navigate to Control and Simulation » Algorithms » "CDEx Model Predictive Control for MIMO.vi". There are a further 3 examples within Control and Simulation » Control Design » MPC.
Hope this helps!
08-18-2011 05:27 AM
Can you please send me that example. I am using LABVIEW 2010 and I dont have that example. Also What Kind of error my model can have??
08-18-2011 05:41 AM
I have attached a zip of a few of the stock examples which should be added when you install the Control Design and Simulation Module. Try navigating to "<root>\Program Files (x86)\National Instruments\LabVIEW 2010\examples\Control and Simulation\Control Design\MPC" - there should be lots more examples there.
I'm not sure what problems there could be with your model. If you are using the System Identification Toolkit then there could be some issues in your code there, could you post that?
08-18-2011 05:49 AM
Thanks for your quick reply. I cant open the VI for MIMO as it in 2011, I have 2010. Also Attached is the VI that I used for my system identification. Just to let you know that my system has one manipulated variable and two measured disrurbances. I have considered my disturbances as inputs when I was doing system indetification. Is that the right way to do it?
08-18-2011 06:14 AM
Sorry about that, the MIMO example is attached for LabVIEW 2010.
I am unsure if you are supposed to consider your disturbances as an input when developing your SS model. I have done my best to read up on the subject, although I have not had any luck!