Hybrid firefly algorithm with a new mechanism of gender distinguishing for global optimization.

Firefly algorithm (FA) is a novel stochastic optimization technique mimicking the inter-attraction and movement of fireflies. As we all know that male and female fireflies exist in nature, which can be distinguished based on some certain definite characteristics. However, all fireflies are unisex in original firefly algorithm (OFA), which is not true biologically. Motivated by the physical characteristics and movement patterns of fireflies, we proposed a novel hybrid firefly algorithm with a new mechanism of gender distinguishing (HFA-GD). In our proposed al-gorithm, the population is divided into two subgroups based on gender, and improved position update formulas of male and female fireflies are proposed to reduce the count of attractions and overcome untimely convergence. Besides, to lessen the chance of FA's untimely convergence, mutation operators of fireflies with gender dis-tinguishing and perturbation jumps of the global best firefly are introduced. Different from some FAs based on gender, male fireflies fly to brighter males to increase their chances of getting attention from other females, and the female flies to the brighter female in our proposed HFA-GD. The performance validity of HFA-GD is tested on 24 2006 Congress on Evolutionary Computation (CEC 2006) constrained optimization problems (COPs) and 28 CEC 2017 COPs, two constrained engineering optimization problems (CEOPs), and two parameter estimation problems. The comparison results of experiment with several meta-heuristic algorithms have demonstrated the effectiveness of HFA-GD.