Metaheuristics and breakdown tolerance in the context of embedded environment

Résumé

Technological developments, along with the emergence of Industry 4.0 allow for new approaches to solve industrial problems, such as the Flexible Job-shop Scheduling Problem (FJSP). The scheduling should deal with a smart and distributed manufacturing system supported by novel and emerging manufacturing technologies such as Cyber- Physics Systems (connected embedded systems). The scheduling research needs to shift its focus to smart distributed scheduling modeling and optimization. In order to transferring traditional scheduling into smart distributed scheduling, we aim to answer these question first: what traditional scheduling methods and optimization algorithms can be combined and reused with connected embedded systems. In this sense, develop an adaptable optimization algorithm into Embedded system is a highly promising approach to solve this problem before a distributed solution. The objective of this thesis is to provide a flexible optimization algorithm, able to make decisions to adapt to changes in the work environment (to communicate with the dynamic environment) and deal promptly with unexpected and internal uncertainties failures. In order to accomplish these goals, we move forward in the following way: • First, we supply an effective solution resolution to the FJSP by utilizing population based metaheuristics (P-metaheuristics) and make a comparison of these algorithms to eventually find out that the edited model of basic PSO (MPSO) might be the most suitable method to resolve problems. An improvement of the run-time of the serial PSO and a simulation-based comparison of P-metaheuristics for FJSP either with or without a fuzzy processing time are done.We are essentially studying sensitivity to the population size and generations number. • Second, we provide the MPSO to solve FJSP: Firstly, we make an analysis and improve the particle coding under the problem (PSO-OMS and PSO-JMS for FJSP are proposed). The results of the experiment prove that PSO-OMS gives the best results in a minimum run time with a guarantee of particles convergence. Optimization of CPU time is taken into account and the general framework is addressed with the five types of constraints which are considered here: 1) the task execution time constraint (MS) 2) the machine load constraint (workload) 3) CPU time 4) the number of particles that reaches the minimum (global best) 5) and the nature of convergence of the particles. • Third, we introduce the FJSP with machine breakdown problem. Then we use the PSO-OMS and enhance this algorithmic performance to solve the FJSP under machine breakdown. The robustness and stability of the rescheduling are guaranteed with the quality of result obtained. • Fourth, a modified version of PSO-OMS called Two-level PSO is proposed for the purpose of preparing a PSO variant for embedded applications. • Finally, we are interested to migrate this implementation (Two-level PSO) to an embedded platform. Then we propose an embedded two-level PSO (E2L-PSO) variant adaptive to a dynamic embedded system under these constraints: 1) the task execution time (MS) 2) the machine load (workload) 3) CPU time 4) memory constraint 5) and feasibility in real time.

Citer ce document

Accès au document

Ce document est hébergé sur son dépôt institutionnel d'origine.

Auteur(s)

Statistiques

Consultations : 2

Téléchargements : 0