A multi-objective Genetic Algorithm scheduling tool for solving complex scheduling problems in the capital goods industry

  1. Lookup NU author(s)
  2. Dr Neo Xie
  3. Dr Pupong Pongcharoen
  4. Professor Christian Hicks
Author(s)Xie WB, Pongcharoen P, Hicks C
Publication type Article
JournalInternational Journal of Production Research
ISSN (print)0020-7543
ISSN (electronic)1366-588X
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
This paper outlines the design and development of a multi-objective Genetic Algorithm scheduling tool (MOGAST) that has been developed for scheduling complex products with deep product structure. The tool aims to produce a set of schedules that provide optimum tradeoffs between delivery performance and inventory. A series of experiments based on a full factorial design were undertaken that used data obtained from a collaborating capital goods company that produced complex products in low volume. Genetic Algorithms include important parameters including: the probabilities of crossover and mutation; the population size; and the number of generations. The crossover and mutation operators and the selection scheme also need to be specified. It is important for multi-objective Genetic Algorithms (MOGAs) to maintain a diverse population and produce uniformly distributed values over a Pareto front. Fitness sharing has been widely used to artificially reduce the fitness of solutions in densely populated areas. The approach requires the user to specify the niche size σshare. The aims were to solve various sizes of industrial scheduling problems and to identify the best configuration of the multiple objective Genetic Algorithm in terms of: i) the probabilities of crossover and mutation; ii) the combination of crossover and mutation operators; iii) selection schemes; and iv) the niche size. The tool was successfully used to optimise an 18 month schedule for a facility that manufactured complex products.
PublisherTaylor & Francis Ltd.
NotesPaper resubmitted on 20th January 2012 in response to R&R
Actions    Link to this publication