How is the Kalman filter estimator implemented in an MPC?

[Vicky Vicky]

I have a physical process with 2 inputs, 4 states and 2 outputs. Two of the states are

measured while the other two needs to be estimated using kalman filter.I have

developed the controller (2 i/p-2 o/p) with an MPC. I can see the 4 states in the implemented 

MPC, but how do I implement a K.filter required to estimate the states of the 2 unmeasured 

states in the physical process. ...help help help... 

Attached is the implemented MPC.vi for comments.

[Vicky Vicky]

Hi house, 

I have been waiting for some suggestions about

the implementation of Kalman filter for an MPC

controller. Please, someone should help me its very

urgent. 

[KF_]

Hi Vicky,

I dont have Labview installed on this laptop, but I have implemented the Labview MPC on a distillation column. You have 2 blocks that are necessary for the MPC, one outside the loop and one inside. You can connect all the necesary settings, like the model, the control horizon, the kalman filter settings,... to the block outside the loop. I think the control for Kalman is called estimation parameters. There you select between no estimation, pole placement and Kalman filter. For the Kalman filter you have to give the controller 3 matrices, that define your Kalman filter, N, Q and R. These matrices are related to your noise model, I don't know the exact theory unfortunately.

regards

KF