A genetic-based solution to the task-based sailor assignment problem
This chapter presents a study investigating a multi-objective formulation of the United States Navy's Task-based Sailor Assignment Problem and examines the performance of a widely used multi-objective evolutionary algorithm (MOEA), namely NSGA-II, on large instances of this problem. The performance of the evolutionary algorithm is examined with respect to both solution quality and diversity and has shown to provide inadequate diversity along the Pareto front. Domain-specific local improvement operators were introduced into the MOEA, producing significant performance increases over the evolutionary algorithm alone. Thus, hybrid MOEAs provided greater diversity along the Pareto front. Also a parallel version of the evolutionary algorithm was implemented. Particularly, an island model implementation was investigated. Exhaustive experimentations of the sequential and parallel implementations were carried out. The experimental results show that the genetic-based solution presented here is suitable for these types of problems.
Variants of Evolutionary Algorithms for Real-World Applications
Dasgupta, D., Garrett, D., Nino, F., Banceanu, A., & Becerra, D. (2012). A genetic-based solution to the task-based sailor assignment problem. Variants of Evolutionary Algorithms for Real-World Applications, 9783642234248, 167-203. https://doi.org/10.1007/978-3-642-23424-8_6