Hello everybody,

In the context of a generalized uncertain LTI plant G with inputs (d,u) and outputs (v,y):

[ A | B1 B2 ]

---------------

[ C1 | D11 D12]

[ C2 | D21 D22],

a common assumption is that D22 (from the command to the measurements) is zero.

But how limiting is to further assume that D21 (from the disturbance to the measurement) is also zero? I have found to be the case in certain mechanical systems (inverted pendulum, hard disk drive, mass-spring-damper etc.), but is it a fair generalization? Can you provide any reference which considers this assumption of D21=0 in any developed research? Or an actual example of a model where D21 is other than zero.

Thank you for your input,

Mircea.