Dynamic Scale Genetic Algorithm: An Enhanced Genetic Search for Discrete Optimization

The minimization of operations and support resources of reusable launch vehicles is a complex task, involving discrete optimization and the simulation domain. Genetic algorithms, offering a robust search strategy suitable for integer variables and the simulation domain, can be applied to minimize these resources. This research developed an enhanced genetic algorithm for problems with a linear objective function, the most common class of discrete optimization problems. The dynamic scale genetic algorithm developed here incorporates concepts of implicit enumeration to enhance search. This is achieved by utilizing problem specific information to refine the solution space over successive generations. The utility of the proposed algorithm was demonstrated by comparing its performance, in terms of quality of solutions produced, to that of the simple genetic algorithm. For all test problems, the dynamic scale genetic algorithm consistently produced better solutions in fewer generations. The proposed algorithm was successfully applied to optimize the operation and support resources of reusable launch vehicles, through a discrete event simulation model. The least cost solution so obtained represents an improvement over both the simple genetic algorithm, and the previous manual approach of minimizing operation and support resources