Conditions under which induced $\mathcal{L}_2$ system gain is equal to induced $\mathcal{L}_\infty$ system gain

54 Views Asked by At

For the following inequalities refer

  • Zhou, Kemin; Doyle, John C.; Glover, Keith, Robust and optimal control, Upper Saddle River, NJ: Prentice Hall. xx, 596 p. (1996). ZBL0999.49500.

Consider a general LTI system

For $u(t) \in \mathcal{L}_2$ the input signal and $y(t) \in \mathcal{L}_2$ the output signal, the induced system gain is given by $\lVert G \rVert_\infty$ as shown below (Ref. Page 107 Table 4.2)

$$ \lVert z \rVert_2 \leq \lVert G \rVert_\infty \lVert u \rVert_2 $$

similarly, for $u(t) \in \mathcal{L}_\infty$ the input signal and $y(t) \in \mathcal{L}_\infty$ the output signal, the induced system gain is given by $\leq \int_0^\infty \lVert g(t \rVert dt$ as shown below (Ref. Page 107 Table 4.2)

$$ \lVert z \rVert_\infty \leq \int_0^\infty \left\lVert g(\tau) \right\rVert d\tau \lVert u \rVert_\infty $$

Also we have the inequality (Ref. Page 111 Theorem 4.5)

$$ \lVert G \rVert_\infty \leq \int_0^\infty \lVert g(t) \rVert dt $$

The question that I want to ask is under what conditions does equality holds for the above stated inequality?

1

There are 1 best solutions below

1
On BEST ANSWER

There is no known necessary conditions for that and it is unlikely that there are any as it can just be a coincidence.

In any way, there is a small class of systems for which this is always true and this is the case of SISO internally positive systems. A system is said to be internally positive if for any nonnegative initial condition and any nonnegative input, the state remains nonnegative and the output is nonnegative. A necessary and sufficient for a system $(A,B,C,D)$ to be internally positive is that $A$ be Metzler, and that the other matrices be nonnegative (i.e. componentwise).

In the case of a SISO stable internally positive system, we have that all the $L_p$-gains are the same and are equal to $G(0)$ where $G(s):=C(sI-A)^{-1}B+D$.