Generalized disjunctive constraint propagation for solving the job shop problem with time lags

Abstract

International audienceIn this paper we propose new insights based on an insertion heuristic and generalized resource constraint propagation for solving the job shop scheduling problem with minimum and maximum time-lags. To show the contribution of our propositions we propose a branch-and-bound algorithm and provide an experimental study. The results obtained conclude that our heuristic obtains feasible schedules with a better makespan than previous approaches, especially for instances with tightened time lags. The results also prove the interest of the constraint propagation generalization when time lags are considered

Solving Variants of the Job Shop Scheduling Problem Through Conflict-Directed Search

“…Caumond et al (2008) introduced a memetic algorithm, which is a combination of population based search and local search with an emphasis on using problem specific knowledge. Artigues et al (2011) proposed a CP branch and bound method using dichotomic search and the branching scheme of Torres and Lopez (2000). to the optimal job insertion problem.…”

Scatter search with path relinking for the job shop with time lags and setup times

“…Artigues et al [23] study the JSPTL as well. They propose a job insertion heuristic for generating initial solutions and generalized resource constraint propagation mechanisms, which are embedded in a branch-and bound algorithm.…”

Weight-based Heuristics for Constraint Satisfaction and Combinatorial Optimization Problems

“…The problem is also studied in [1] by integrating generalized resource constraint propagation and branch-and-bound.…”

“…In this part, as in [1,6], we only consider maximum time-lags between successive operations of the same job. In this case, a trivial schedule can be obtained by a greedy constructive algorithm.…”

“…To solve the JSPTL under study, we use the job insertion heuristic proposed by [1]. It consists in selecting a job J i and in setting the start time of its first operation, t i,1 , to its earliest start time on the associated machine.…”