Using SOS for Analysis of Zeno Stability in Hybrid systems with Nonlinearity and Uncertainy

Abstract: Hybrid systems exhibit phenomena which do not occur in systems with continuous vector fields. One such phenomenon - Zeno executions - is characterized by an infinite number of discrete events or transitions occurring over a finite interval of time. This phenomenon is not necessarily undesirable and may indeed be used to capture physical phenomena. In this paper, we examine the problem of proving the existence and stability of zero executions. Our approach is to develop a polynomial-time algorithm - based on the sum-of-squares methodology - for verifying the stability of a Zeno execution. We begin by stating Lyapunov-like theorems for local Zeno stability based on existing results. Then, for hybrid systems with polynomial vector fields, we use polynomial Lyapunov functions and semialgebraic geometry (Positivstellensatz results) to reduce the local Lyapunov-like conditions to a convex feasibility problem in polynomial variables. The feasibility problem is then tested using an algorithm for sum-of-squares programming - SOSTOOLS. We also extend these results to hybrid system with parametric uncertainty, where the uncertain parameters lie in a semialgebriac set. We also provide several examples illustrating the use of our technique.