On the optimal control of hybrid systems: Analysis and zonal algorithms for trajectory and schedule optimization

In [10], [11], [12] a class of hybrid optimal control problems was formulated and a set of necessary conditions for hybrid system trajectory optimality was presented. Employing these conditions, we presented and analyzed a class of general Hybrid Maximum Principle (HMP) based algorithms for hybrid systems optimization. In this paper it is first shown how the HMP algorithm class can be extended with discrete search algorithms which find locally optimal switching schedules and their associated switching times. We then present the notion of optimality zones; these zones have a well defined geometrical structure and once they have been computed (or approximated) they permit the exponential complexity search for optimal schedule sequences of the first method to be reduced to a complexity level which under reasonable hypotheses is proportional to the number of zones. The algorithm HMP[Z] which performs this optimization is essentially a minor modification of the HMP algorithm and permits one to reach the global optimum in a single run of the HMP[MCS] (see [11], [12]) algorithm. The efficacy of the proposed algorithms is illustrated via computational examples.