Does the separation principle hold for a model predictive controller, assuming all internal models are linear?

29 Views Asked by At

The question says it all, I think. Say I have an MPC and a plant that has at least one unobserved state. Can I create an observer for that state and feed the observed system's outputs into the MPC, and use the separation principle to design that observer without consideration of the MPC, presuming my goal is system stability?