Symbolic Analysis of Hybrid Systems Involving Numerous Discrete Changes Using Loop Detection

Hybrid systems are dynamical systems that include both continuous and discrete changes. Some hybrid systems involve a large or infinite number of discrete changes within an infinitesimal-width region of phase space. Systems with sliding mode are typical examples of such hybrid systems. It is difficult to analyze such hybrid systems through ordinary numerical simulation, since the time required for simulation increases in proportion to the number of discrete changes. In this paper, we propose a method to symbolically analyze such models involving numerous discrete changes by detecting loops and checking loop invariants of the model's behavior. The method handles parameterized hybrid systems and checks inclusion of parameterized states focusing on the values of a switching function that dominate the dynamics of sliding mode. We implemented the main part of the method in our symbolic hybrid system simulator HyLaGI, and conducted analysis of example models.