Job-shop scheduling by combining reachability analysis with linear programming

Abstract For optimization problems in planning and scheduling, techniques based on timed automata (TA) have the advantage of a simple and intuitive problem specification. In addition, search algorithms operating on the reachability tree of the TA have been proposed for determining the optimal (or a near-optimal) model evolution relatively efficiently. In order to improve the efficiency of the solution further, this paper describes the extension of a recently developed approach in which the search tree is pruned based on lower bounds for the cost that are computed by embedded linear programming. The paper shows how the use of a tailored linear program can considerably reduce the effort to compute the lower bounds such that the overall performance significantly increases.