A Novel Adaptive Bat Algorithm to Control Explorations and Exploitations for Continuous Optimization Problems

Swarm intelligence (SI) algorithms generally come from nature or biological behavior of nature. These algorithms use probabilistic search methods that simulate the behavior of biological entities or the natural biological evolution. Swarm intelligence (SI) is based on collective behavior of selforganized systems. Typical swarm intelligence algorithms include Particle Swarm Optimization (PSO), Ant Colony System (ACS), Bacteria Foraging (BF), the Artificial Bee Colony (ABC), and so on. Recently some new swarm based algorithms like Firefly Algorithm (FA) and Bat Algorithm (BA) has emerged. BA is a new optimization technique, which is based on the echolocation behavior of bats. BA is very efficient in exploitations but relatively poor in explorations. In this paper, a Novel Adaptive Bat Algorithm (NABA) is presented to improve the explorative characteristics of BA. The proposed algorithm incorporates two techniques within BA to improve its degree of explorations, which include the Rechenberg’s 1/5 mutation rule and the Gaussian probability distribution to produce mutation step sizes. Both these techniques try to balance between the explorative and exploitative properties of BA. Simulation results on a number of benchmark functions on the continuous optimization problem suggest that the proposed algorithm – NABA often show much improved results, compared to the standard BA.