Multi-objective grey wolf optimizer based on decomposition.

Highlights • An enhanced algorithm for solving multi-objective optimization problems is proposed. • The proposed algorithm improves the original Multi-Objective Grey Wolf Optimizer. • This improvement is related to the decomposition of multi-objective problems. • The proposed approach is evaluated on benchmark functions and real-world applications. • Statistical analysis corroborates the good performance of the proposed algorithm. Abstract An optimization technique aims to find the best solution to an optimization problem. If the problem considers only one objective function, the best solution will provide the optimal value for such objective. However, if the problem considers two or more objectives, the selection of solutions will not be that straightforward since such objective functions are usually in conflict. For this kind of optimization problems, the use of analytical or exact methods becomes impractical. Thus, heuristic or metaheuristic approaches have to be applied for finding the optimal solutions or, at least, approximate solutions to the optimum. As a consequence, a wide variety of metaheuristics inspired by nature has been proposed for solving optimization problems. Among them, the Grey Wolf Optimizer is a metaheuristic of recent creation that in the last few years has attracted the attention of many researchers. Furthermore, a multi-objective extension of this technique was recently introduced proving its high performance comparable to other multi-objective optimization methods. In this paper, a multi-objective grey wolf optimizer based on the decomposition is introduced. Our proposed algorithm approximates Pareto solutions cooperatively by defining a neighborhood relation among the scalarizing subproblems in which the multi-objective problem is decomposed. The performance of our proposed method is compared against those achieved by six state-of-the-art bio-inspired techniques showing its high performance in both, well-known benchmark problems and two real-life engineering problems. [ABSTRACT FROM AUTHOR]