Evolutionäre Verfahren zur Optimierung von Produktionsplänen mittels implizierter Kooperation

This work describes a new approach to cooperative production planning and control using genetic algorithm and timed hierachical object-related Petri Nets. Every planning problem is decomposed into subproblems like scheduling, lotsizing etc. Each subproblem is solved independently of the other subproblems with an individual genetic algorithm. The individual solutions are then synthesized into a global production plan. This leads to a pool of new solutions. Every global production plan is mapped to object tokens. The fitness of the object tokens is calculated using a Petri Net-based architecture for plant simulation. Every feasible production plan is the solution of a four-objective problem with conflicting objective functions. The genetic algorithms are modified to deal with multiple objectives by incorporating the concepts of Pareto domination and goal programming. The fitness of the global solution controls the direction of search taken by every individual genetic algorithm. Because there is no explicit communication between the individual genetic algorithms, they cooperate implicitly. Experimental results show that the combination of well known heuristics with implicit cooperation outperforms well known heuristics without implicit cooperation.

Grantor dc:publisher

Publikationsserver der RWTH Aachen University

Year dc:date

2003

Author dc:creator

• Boll, Helmut

Contributors dc:contributor

• Sonnenschein, Michael

• info:eu-repo/classification/ddc/004
• Informatik
• evolutionäre Algorithmen
• Produktionsplanung
• Petrinetze

Statement dc:rights

• info:eu-repo/semantics/openAccess

Language dc:language

ger

Identifier

https://publications.rwth-aachen.de/search?p=id:%22RWTH-CONV-123579%22

OAI identifier oai:identifier

oai:publications.rwth-aachen.de:61977