No abstract

In this paper we tackle a variant of the job shop scheduling problem where task durations are uncertain and only an interval of possible values for each task duration is known. We propose a genetic algorithm to minimise the schedule's makespan that takes into account the problem's uncertainty during the search process. The behaviour of the algorithm is experimentally evaluated and compared with other stateof-the-art algorithms. Further analysis in terms of solution robustness proves the advantage of taking into account interval uncertainty during the search process with respect to considering only the expected processing times and solving the problem's crisp counterpart. This robustness analysis also illustrates the relevance of the interval ranking method used to compare schedules during the search.

In this paper, a novel Artificial Bee Colony algorithm is proposed to solve a variant of the Job Shop Scheduling Problem where only an interval of possible processing times is known for each operation. The solving method incorporates a diversification strategy based on the seasonal behaviour of bees. That is, the bees tend to explore more at the beginning of the search (spring) and be more conservative towards the end (summer to winter). This new strategy helps the algorithm avoid premature convergence, which appeared to be an issue in previous papers tackling the same problem. A thorough parametric analysis is conducted and a comparison of different seasonal models is performed on a set of benchmark instances from the literature. The results illustrate the benefit of using the new strategy, improving the performance of previous ABC-based methods for the same problem. An additional study is conducted to assess the robustness of the solutions obtained under different ranking operators, together with a sensitivity analysis to compare the effect that different levels of uncertainty have on the solutions’ robustness.

This paper addresses a variant of the job shop scheduling problem with total tardiness minimization where task durations and due dates are uncertain. This uncertainty is modelled with intervals. Different ranking methods for intervals are considered and embedded into a genetic algorithm. A new robustness measure is proposed to compare the different ranking methods and assess their capacity to predict ‘expected delays’ of jobs. Experimental results show that dealing with uncertainty during the optimization process yields more robust solutions. A sensitivity analysis also shows that the robustness of the solutions given by the solving method increases when the uncertainty grows.

No abstract