Solving Optimization Problems on Hybrid Systems by Graph Exploration

Abstract

The aim of this research is to give a systematic way to solve optimization problems on hybrid systems. The target classes of hybrid systems are linear hybrid automata and piecewise linear systems. The behavior of a hybrid system is represented by a directed acyclic graph such that each node is associated with a convex polyhedron on continuous state variables, and each edge corresponds to a discrete transition. Then the problem is solved by finding a trajectory in the graph that fulfills given specifications and also optimizes a given objective function. The procedure is implemented on a constraint logic programming language