Evolutionary algorithms are generally based on populations of solutions which are subject to the application of operators such as recombination, mutation, and selection in order to evolve the population and eventually obtain high-quality solutions. Different, yet often similar evolutionary algorithms are discussed in various research communities from different perspectives. In this work we strive for a better understanding of the performance of different designs within the general framework of evolutionary computation. We examine and compare (discrete) particle swarm optimization with classic genetic algorithms, both with and without hybridization with local search. In particular, we analyze the effect of different selection and reproduction mechanisms on solution quality, population diversity, and convergence behavior, and examine approaches for maintaining population diversity. As application we consider an NP-hard combinatorial optimization problem, namely the no-wait (continuous) flow-shop scheduling problem with flow-time criterion. The computational results support the importance of local search within (hybridized) evolutionary algorithms and show how solution quality depends on a reasonable design of crossover operators, distance functions, population diversity measures, and the control of population diversity. © Springer-Verlag Berlin Heidelberg 2009.
CITATION STYLE
Czogalla, J., & Fink, A. (2009). Design and analysis of evolutionary algorithms for the no-wait flow-shop scheduling problem. Lecture Notes in Economics and Mathematical Systems, 624, 99–126. https://doi.org/10.1007/978-3-642-00939-6_7
Mendeley helps you to discover research relevant for your work.