H2 control is try to find K to minimize the transfer function from w -> z.
LQG is try to find kalman gain K to minimize the steady-state covariance of the error.
What is the same things and different things between them?
When I read some papers, it goes like the optimal H2 (or LQG) control ..... So what type of problems in common is for LQG and H2 and what type of problems is for one but not for the other?
Thanks
LQR, LQG and LQE are used when all the states are measurable or update-able. In LQG if the states are not measurable directly, they are first estimated and by the separation principle a state feedback controller can be designed. In H2 problems one can select special performance channels and use partial state information. No need to measure or estimate all the states. H2 synthesis can be used for output feedback control.