Design and development of a real-time scheduling system in a sensorised job shop using cloud-based simulation with mobile device access

Snyman, Stephan (2019-12)

Thesis (PhD)--Stellenbosch University, 2019.

Thesis

ENGLISH ABSTRACT: Traditional scheduling approaches used in manufacturing systems address scheduling problems before production commences, which poses possible problems when the system is interrupted by unexpected events. Managers must then react in a timely fashion, by developing a new or revised schedule to mitigate the effects of the interruptions on the productivity of the system. This can be done by a real-time scheduling system, which is used in conjunction with the actual manufacturing system. Technological advances, including cloud-based computing, the omnipresence of mobile devices, and the improved capabilities of sensor networks, have opened up the opportunity to design a real-time scheduling system, as well as create software architectures to support such a system. The purpose of this research project is therefore to develop a prototype of a real-time simulation scheduling system, which will serve as a decision support tool for real-time rescheduling of machine steps in a job shop. The prototype incorporates a cloud-based information system for the storage of data and a cloud-based simulation scheduler that generates schedules. It also includes web pages for logging data changes and selecting a new schedule and sensors that keep track of the movement of jobs through the job shop. The preliminary test results of the developed simulation scheduler suggest that metaheuristics should be considered to generate schedules, due to the metaheuristics outperforming the common dispatching rules. The model was then expanded from the single-objective to the multi-objective domain, which is a better representation of the real-world job shop environment. Several metaheuristics were adapted to solve the bi-objective job shop scheduling problem, after which comparison tests were conducted. The tests revealed that the NSGAII performed best of all the metaheuristics and it was selected for further implementation. The final phase of this research project was to implement a newly developed ranking and selection procedure for discrete stochastic simulation problems, called MMY. The MMY procedure finds the minimum number of simulation replications for each solution, while guaranteeing that the probability of correct selection of the best solutions exceeds a desired value. In this study, MMY finds the best simulated schedules while the probability of correct selection is guaranteed.

AFRIKAANSE OPSOMMING: Tradisionele skeduleringsbenaderings wat gebruik word in vervaardigingstelsels spreek skeduleringsprobleme, voordat produksie begin, aan. Dit kan moontlike probleme veroorsaak wanneer die stelsel onderbreek word deur onverwagte gebeurtenisse. Bestuurders moet dan betyds reageer deur 'n nuwe of hersiene skedule te ontwikkel om sodoende die uitwerking van die onderbrekings op die produktiwiteit van die stelsel te versag. Dit kan gedoen word deur 'n reële tyd skeduleringstelsel, wat gebruik word in samewerking met die werklike vervaardigingstelsel. Baie tegnologiese vooruitgange, insluitende wolkgebaseerde rekenaars, die alomteenwoordigheid van mobiele toestelle en die verbeterde vermoëns van sensornetwerke, het die geleentheid gebied om die reële tyd skeduleringstelsel te ontwerp, asook die geleentheid vir die skep van sagtewareargitekture om hierdie skeduleringstelsels te ondersteun. Die doel van hierdie navorsingsprojek is dus om 'n prototipe van 'n re ele tyd simulasie skeduleringstelsel te ontwikkel, wat sal dien as 'n besluit hulpmiddel vir die re ele tyd herskedulering van masjienstappe in 'n werkswinkel. Die prototipe bevat 'n wolkgebaseerde inligtingstelsel vir die stoor van data asook 'n wolkgebaseerde simulasieskeduleerder wat skedules genereer. Dit sluit ook webbladsye in om dataveranderinge aan te teken en sensors wat die beweging van werk deur die werkswinkel volg. Die voorlopige toetsuitslae van die ontwikkelde simulasie skeduleerder stel voor dat metaheuristieke oorweeg moet word om skedules te genereer, weens die metaheuristieke wat beter as die algemene versendingsre els presteer. Die model kan dan uitgebrei word van die enkel-doelwit tot die multi-doelwit domein, wat 'n beter voorstelling van die werklike werkswinkelomgewing is. Verskeie metaheuristieke is aangepas om die bi-doelwit werkswinkel skeduleringsprobleem op te los, waarna vergelykingstoetse uitgevoer kon word. Die uitkoms van die toetse het getoon dat die NSGAII die beste presteer het van al die metaheuristieke en daarom is dit gekies vir verdere implementering. Die nale fase van hierdie navorsingsprojek was om 'n nuut ontwikkelde rangskiken- kies prosedure vir diskrete stogastiese simulasieprobleme, genaamd MMY, te implementeer. Die MMY prosedure vind die minimum aantal simulasie replikasies vir elke oplossing, terwyl dit die waarskynlikheid van korrekte keuse waarborg. In hierdie studie vind die MMY prosedure die beste gesimuleerde skedules terwyl die waarskynlikheid van korrekte keuse waarborg word.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/107020
This item appears in the following collections: