A Novel Dynamic Multi-objective Evolutionary Algorithm with an Adaptable Roulette for the Selection of Operators

In this chapter, the optimization of Dynamic Multi-Objective Problems (DMOP) is approached. To solve this kind of problems several evolutionary algorithms with a static selection of operators are reported in the literature. In this work, a new evolutionary algorithm with that an online operator selector is proposed. The operator choice is guided by a self-adapting roulette that modifies the probabilities of usage for each operator. The evolutionary algorithm proposed follows the classical generational scheme of an evolutionary algorithm, but each offspring is constructed by selecting an operator from an operator’s pool based on a probability regulated by the roulette. A series of experiments were done to assess the performance of the proposed algorithm that includes a set of state-of-the-art algorithms, a set of standard instances and statistical hypothesis tests to support the conclusions.