E. Hart et al., Scheduling chicken catching - An investigation into the success of a genetic algorithm on a real-world scheduling problem, ANN OPER R, 92, 1999, pp. 363-380
Genetic Algorithms (GAs) are a class of evolutionary algorithms that have b
een successfully applied to scheduling problems, in particular job-shop and
flow-shop type problems where a number of theoretical benchmarks exist. Th
is work applies a genetic algorithm to a real-world, heavily constrained sc
heduling problem of a local chicken factory, where there is no benchmark so
lution, but real-life needs to produce sensible and adaptable schedules in
a short space of time. The results show that the GA can successfully produc
e daily schedules in minutes, similar to those currently produced by hand b
y a single expert in several days, and furthermore improve certain aspects
of the current schedules. We explore the success of using a GA to evolve a
strategy for producing a solution, rather than evolving the solution itself
, and find that this method provides the most flexible approach. This metho
d can produce robust schedules for all the cases presented to it. The algor
ithm itself is a compromise between an indirect and direct representation.
We conclude with a discussion on the suitability of the genetic algorithm a
s an approach to this type of problem.