Scheduling of deadlock and failure-prone automated manufacturing systems via hybrid heuristic search

Document Type

Article

Publication Date

6-3-2017

Abstract

This work focuses on the scheduling problem of deadlock and failure-prone automated manufacturing systems, and presents a new scheduling method by combining a robust supervisory control policy and hybrid heuristic search. It aims to minimise makespan, i.e. the completion time of the last part. Based on the extended reach ability graph of the system, it establishes a new heuristic function and two dispatching rules to guide the search process for a schedule. By embedding a robust supervisory control policy into the search process, it develops a polynomial robust dynamic window search algorithm. Failure and repair events of unreliable resources may occur during the execution of a schedule obtained by the proposed algorithm and may make the schedule infeasible. To reduce the influence caused by them and ensure all parts to be finished, this work proposes two event-driven strategies. The first one suspends the execution of the parts requiring failed resources and those to be started until all failed resources are repaired and permits only those parts that have already been processed on working machines to be completed. The second one invokes the proposed algorithm to obtain a new schedule at the vertex generated after a resource failure or repair event and executes the new schedule. Both strategies are effective while the latter performs better at the expense of more computation.

Identifier

85016125963 (Scopus)

Publication Title

International Journal of Production Research

External Full Text Location

https://doi.org/10.1080/00207543.2017.1306132

e-ISSN

1366588X

ISSN

00207543

First Page

3283

Last Page

3293

Issue

11

Volume

55

This document is currently not available here.

Share

COinS