Autonomous Particles Groups for Synchronous-Asynchronous Particle Swarm Optimization

In recent years several variations of Particle Swarm optimizations have been proposed due to their simplicity and powerful global search ability. The Synchronous-Asynchronous Particle Swarm Optimization (SAPSO) algorithm offers a modification from the original PSO that divides the particles into smaller groups. The SAPSO presents an asynchronous update mechanism in terms of exploration and a synchronous update mechanism in terms of exploitation. The SAPSO present a lack of exploitation intensification at the last iteration. In order to overcome this deficiency, is proposed a modified SAPSO algorithm called Autonomous Groups Synchronous-Asynchronous Particle Swarm Optimization (AG-SAPSO). Exploitation and exploration rate of SAPSO is improved through a decrescent function that determines the number of groups employed in the search process in a fixed population and rearranging of Groups in each iteration. Nine widely known benchmark functions were utilized to evaluate the yield from AG-SAPSO that includes unimodal and multimodal type, also rotate and shift features. Experimental results demonstrate that the AG-SAPSO has a better yield for most functions, in contrast with other PSO modifications and nature-inspired optimization algorithms.