Gravitational search algorithm (GSA) is a recent population based metaheuristics algorithm. It has a good ability to perform a wide exploration and a deep exploitation, however it becomes inactive when the premature convergence happens and loses its ability to explore new solutions in the search space. In order to avoid this problem, we propose in this paper a new hybrid gravitational search algorithm with a L´evy flight operator. The proposed algorithm is called Hybrid Gravitational Search with L´evy Flight (HGSLF).When the distance between two masses become very close and both of them are not a best solution in the population, the L´evy flight operator is applied on one of them to increase the diversity of the algorithm and avoid trapping in local minima. The general performance of the proposed HGSLF algorithm is tested on 13 unconstrained (7 uni-model problems and 6 multi-model problems), 8 constrained optimization problems and compared against 8 different algorithms. The numerical results show that the proposed HGSLF algorithm can solve unconstrained, constrained optimization problems in reasonable time and faster than standard GSA and other comparative algorithms.
Fouad Ali, Ahmed
"A Hybrid Gravitational Search with L´evy Flight for Global Numerical Optimization,"
Information Sciences Letters: Vol. 4
, Article 4.
Available at: https://digitalcommons.aaru.edu.jo/isl/vol4/iss2/4