6. Conclusions
In this paper, a new formulation of the MGS problem was developed to conceive a GA that revolves around RG-MGS, a new randomized greedy method being able to construct, in an intelligent, diversified feasible solutions for this highly constrained problem. This heuristic procedure is employed as initialization mechanism for the GA and it is the basic pillar on which the design of the crossover operators is supported. A design that was completed precisely through the knowledge acquired in the research field of GAs for continuous optimization problems [30,38,39,29]. The GA also integrates a restart operator to ensure a reliable evolution toward promising areas throughout the entire search. The proposal has proven to be a very high performing algorithm for the MGS problem, showing it to be very competitive with respect to state-of-the-art algorithms. Specifically, the empirical study reveals a clear superiority when tackling hard and large instances. The ability of the proposed GA to yield superior outcomes along with the simplicity and flexibility of this approach, allows us to conclude thatthis metaheuristic arises as a tool of choice to face this problem. Moreover, itinvites further consideration to explore other forms of evolutionary algorithms, such as the memetic algorithms [40,39], which apply a local search method to members of the GA populationafter crossover andmutationoperations, withthe aimof exploiting the best search regions identified by the global sampling done by the GA.
