Performance Evaluation of Vector Evaluated Gravitational Search Algorithms Based on ZDT Test Functions

This paper presents a performance evaluation of Vector Evaluated Gravitational Search Algorithm (VEGSA), namely VEGSA-I and VEGSA-II algorithms, for multi-objective optimization problems. The VEGSA algorithms use a number of populations of particles. In particular, a population of particles corresponds to one objective function to be minimized or maximized. Simultaneous minimization or maximization of every objective function is realized by exchanging a variable between populations. Performance evaluation is done based on ZDT test functions, which is a common benchmark problem for multiobjective optimization. The results shows that both VEGSA algorithms are outperformed by other multi-objective optimization algorithms and further enhancements are needed before it can be employed in any application. Keywords-gravitational search algorithm; multi-objective optimization problem;