Properties of behavioral models for a class of hybrid dynamical systems

A class of models developed for online monitoring of complex processes which exhibit both continuous and discrete dynamics is presented. The authors describe the basic elements of the hybrid system modeling formalism and the trajectory encoding method for maintaining an online representation of the set of valued system trajectories that are consistent with the observations from the process. It is shown analytically that the class of models and associated observation relations defined satisfy two properties essential for a modeling formalism to be used in online monitoring, i.e. causality and time monotonicity. A brief description is presented of current research into extensions of the trajectory encoding method to monitor systems with divergent dynamic evolutions