Solving the Multi-objective Flexible Job-Shop Scheduling Problem with Alternative Recipes for a Chemical Production Process

Abstract

This paper considers a new variant of a multi-objective flexible job-shop scheduling problem, featuring multisubset selection of manufactured recipes. We propose a novel associated chromosome encoding and customise the classic MOEA/D multi-objective genetic algorithm with new genetic operators. The applicability of the proposed approach is evaluated experimentally and showed to outperform typical multiobjective genetic algorithms. The problem variant is motivated by realworld manufacturing in a chemical plant and is applicable to other plants that manufacture goods using alternative recipes.

Cloud-Based Integrated Process Planning and Scheduling Optimisation via Asynchronous Islands

“…The considered optimisation problem is an example of a typical manufacturing planning and scheduling problem that can be found in any discrete manufacturing company. However, the used optimisation engine can be also applied to process manufacturing as described in [1]. We first evaluate the communication overhead of the proposed cloud architecture by measuring the response time of push and pull operations.…”

“…The Diversity Comparator Indicator (DCI) [18] has been used as the quality indicator CI for measuring the quality of Pareto Front approximations obtained in the subsequent execution stages. DCI quantifies the diversity of the given approximation similarly as in [1]. Five asynchronous islands have been created in the Kubernetes cluster.…”

“…Firstly, they are notorious to require substantial computation and secondly, as the optimisation process is triggered when a smart factory state changes, the needs for huge computational resources are interleaved with idle intervals. Yet the problem of cloud-based optimisation of realistic industrial problems is relatively unpopular in academia [1]. One of the reasons is the innate heterogeneity of the manufacturing process, as discussed, e.g., in [2], and hence the difficulties in proposing an optimisation framework generic enough to be applicable to a wide range of manufacturing optimisation problems.…”

Cloud-based dynamic distributed optimisation of integrated process planning and scheduling in smart factories

“…The innate barrier imposed after each stage is planned to be lifted by proposing a fully distributed algorithm for the subpopulation evolution in the islands. Finally, customised islands management model and GA operators will be proposed to further improve the optimisation results, similarly as in [5].…”

Process planning and scheduling optimisation with alternative recipes

“…Some examples of them are Flexible Jobshop Scheduling Problem (FJSP), where each job can be processed on any machine, FJSP with process plan exibility (FJSP-PPF) [12] or JSP with alternative process plans [13]. Recently, a multi-objective Flexible Job-Shop Scheduling Problem with Alternative Recipes (FJSPAR) has been introduced in [4]. In FJSPAR, each bulk commodity can be produced in various quantities using di er e ntr eci pes with di er e n t m anuf act uri n gti m e , e ner g ycons u mpti on etc.…”