Comparison Of Firefly Photinus Algorithm With Tidal Force Firefly Algorithm For Global Optimisation

Abstract

- Particle Swarm Optimization (PSO) is a widely used metaheuristics algorithm based on the
collective movement of birds or such species. Firefly algorithm (FA) is a modification of conventional
PSO and getting high acceptance in engineering problems due to faster convergence and robustness.
But there is a serious drawback of getting trapped into local optima with FA. A balance between
exploitation and exploration is necessary for avoiding such a situation. This paper investigates further
modification to FA to avoid this limitation. Two such modifications are considered for examination,
Firefly Photinus Algorithm (FPA) and Tidal Force Firefly Algorithm (TFFA). FPA is based on the
behaviour of Photinus fireflies and is developed by incorporating a forbidden mate list into the
algorithm. A time-dependent absorption coefficient is also introduced. In TFFA the firefly attraction
is modified by the tidal force equation. The concept of opposition-based Reinforcement Learning is
also included in the algorithm for minimizing the bias due to the initial selection of the
population. Four standard benchmark optimization functions are used for testing the algorithms. Both
FPA and TFFA give good convergence with TTFA having an edge over the other.