Dynamic Virtual Bats Algorithm (DVBA) for Global Numerical Optimization

This paper presents a novel Dynamic Virtual Bats Algorithm (DVBA) for global optimization. This algorithm is inspired by bat´s echolocation behavior, in particular, focusing on the way they change the wavelength and frequency of the emitted sound wave while looking for prey. The role based search is developed to improve global and local search capability of Yang´s Bat Algorithm. In the DVBA, there are just two bats which are dynamically switching roles from the explorer bat to the exploiter bat according to their position. DVBA has been evaluated, in comparison with standard Particle Swarm Optimization (PSO) and standard Bat Algorithm (BA) on a number of mathematical benchmark functions. Experimental results show that the DVBA can provide superior performance than BA and PSO in optimizing these benchmark functions, mainly, in terms of its accuracy and robustness.