Estimating the domain of attraction for uncertain polynomial systems (Q1765187)

The problem of estimating the robust domain of attraction is studied for a nonlinear system of the form \[ \dot x= f(x,\phi), \] where \(x\in\mathbb{R}^n\), \(\phi\) is a vector of uncertain parameter, \(\phi\in\Phi\), and \(\Phi\) is a polytope of \(\mathbb{R}^z\). The map \(f\) is assumed to be a polynomial with respect to both \(x\) and \(\phi\). The approach relies on polynomial Lyapunov functions, whose coefficients may be either constant (when the family of systems admits a common Lyapunov function) or polynomially dependent on \(\phi\). The robust domain of attraction is defined as the intersection of the domains of attraction of each individual system, as \(\phi\) varies on \(\Phi\). In order to estimate it, the author computes lower bounds for the level sets of the Lyapunov functions contained in the region where their derivatives are negative. This is done by an appropriate transformation of the system and by solving a linear matrix inequality optimization problem and an eigenvalue problem.