translating necessary conditions of stability under additional external inputs

63 Views Asked by At

Not sure if title phrases my question correctly. Suppose we are given necessary asymptotic stability conditions of a 2D system at the origin \begin{equation} \dot{x}(t) = f(x(t), u(t))\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;(1)\end{equation} where $x(t):=(x_1(t),x_2(t)) \in U\underset{open}\subset\mathbb{R}^2$ is the state, and $u(t) \in \mathbb{R}$ is an external input. The planar system is actually in the form of $$ \begin{aligned}x_1(t) &= g(x_1(t),x_2(t),u(t)) \\ x_2(t) &= h(x_1(t),u(t))\end{aligned}.$$

Now, let system dynamics be augmented by an auxiliary strictly-increasing time-varying input $\phi : [0,\infty)\to\mathbb{R}^+$ as $$ \begin{aligned}x_1(t) &= g(x_1(t),x_2(t),u(t)) \\ x_2(t) &= H(\phi(t),x_1(t),u(t))\end{aligned}.\;\;\;\;\;\;\;\;\;(2)$$ such that $\phi(t)$ is not a necessary signal for stability, that is, there is no function that renders origin of (2) stable when $H$ does not depend on either $x_1(t)$ or $u(t)$. I was wondering if set of necessary conditions for stability of (1) can be translated to necessary conditions for (2). Should I treat (2) as a time varying perturbation to nominal system (1) or would this be too conservative to compromise the translation of necessary conditions?

Special cases of interest are

  1. $\phi(t):=\phi(0)+\int_0^t u(\tau)d\tau$

  2. $H$ is periodic in $\phi$ (may apply some kind of averaging to study this case?)

As an example to motivate the question, consider an example stability condition for system (1) as $h(x_1(t),u(t)) <= \alpha(x_1(t)u(t))$ $\forall x_1(t)$ in the open neighborhood of origin, $\forall u(t)$, and some class K function $\alpha$. The augmented system does not have to satisfy these conditions $\forall \phi(t)$ but my guts feelings still tell me that they must be satisfied for most of $\phi(t)$.