Howling mechanism based grey wolf optimizer

Abstract

Grey wolf optimizer (GWO) is an efficient optimization approach in the generation of swarm intelligence based techniques. GWO algorithm relies on the leadership quality and hunting mechanism shown by grey wolves. Half of the iteration in GWO are dedicated to exploration and the rest half are used for exploitation. This article presents a modified GWO approach, known as Howling mechanism based grey wolf optimizer (HGWO). In the modified approach two new phases are added namely, “Howling Phase” and “Re-hunting Phase”. In Howling Phase, the solutions are updated based upon their probable values which depends upon the fitness function. The solutions with higher fitness value are assigned higher probability values so higher fit solutions will be given more chances to update their positions. Further, to overcome the problem of stagnation, re-hunting phase is annexed to re-initialize the Alpha (first fittest solution), Beta (second fit solution), and Delta (third fit solution), if they are not updating their positions upto a predetermined limit. To validate the performance of HGWO, 10 benchmark functions are considered and compared with other optimization algorithms such as GWO, Gravitational Search Algorithm (GSA), and Shuffled frog-leaping algorithm (SFLA). The obtained results show the clear supremacy of the proposed HGWO algorithm.