A relational modeling framework is presented for representing a class
of uncertain hybrid dynamical systems exhibiting delayed, infrequent,
and uncertain observations. The model defines ranges of possible state
evolution, and consequently the system trajectories corresponding to
a given observation sequence are typically nonunique. For monitoring a
nd state estimation, it is infeasible to individually track each of th
e potentially infinite number of possible trajectories. This paper pre
sents a method of encoding a family of trajectories in a graphical rep
resentation. Necessary and sufficient conditions are given for the exi
stence of a finite encoding of system trajectories corresponding to re
ceived observation sequences.