A Memetic Algorithm for Due-Date Satisfaction in Fuzzy Job Shop Scheduling

Satisfying flexible due dates in fuzzy job shop by means of hybrid evolutionary algorithms1

“…For problems combining fuzzy durations and fuzzy due dates, the term agreement index was coined in [69] to refer to the degree to which a job's fuzzy completion time satisfies the flexible due-date, and a genetic algorithm was proposed to maximise the minimum agreement index across all jobs. Maximising the minimum agreement index is also the objective of a random-key genetic algorithm in [44], a scatter search method in [27], a hybrid discrete imperialist competition algorithm in [79] and a memetic algorithm in [61]. This memetic algorithm is also applied to maximise the average minimum index, which is also the objective of the co-evolutive method from [82], here for a fuzzy job shop with multi-process routes, and of the multiobjective genetic algorithm from [29], which also attempts to minimise the number of tardy jobs.…”

“…Maximisation of average or minimum agreement index is also one of the objectives of several multiobjective approaches together with makespan minimisation: based on fuzzy decision making using genetic algorithms [33,68], based on lexicographical goal programming also with a genetic algorithm [32] or Paretofront approximation using a genetic algorithm [54], Pareto archive particle swarm optimisation [43] or a memetic algorithm for multi-process routes [76]. For a classical benchmark from [68], the most competitive methods are the random-key genetic algorithm from [44] and the memetic algorithm from [61], while [32] reports minimum and average agreement index values for another set of instances. In fact, these methods obtain almost full due-date satisfaction, suggesting there is no room for improvement in this area.…”

“…In fact, these methods obtain almost full due-date satisfaction, suggesting there is no room for improvement in this area. However, in [61] new results on harder instances make it clear that the problem of due-date satisfaction is far from being solved.…”

“…Section 3 provides a new insight into the two objective functions usually considered in the literature by situating them in the more general framework of fuzzy multicriteria aggregation and decision making and develops the concept of solution robustness regarding due dates and measures thereof. Then, Section 4 enhances the hybrid algorithm from [61], combining a genetic algorithm with local search, to solve the problem. Different genetic operators are considered and, based on the framework of fuzzy multicriteria aggregation, a new heuristic strategy to guide the search through plateaus in the fitness landscape is proposed for the case of the most demanding objective function.…”

“…The goal is to maximise the overall due-date satisfaction, which translates into two different objective functions depending on how this "overall" satisfaction is defined. The proposal builds upon the contribution presented at IWINAC'2017 Conference [61], where a hybrid algorithm for the FJSP was introduced. This preliminary work is enhanced in several ways.…”