Behavior Changing Schedules for Heterogeneous Particle Swarms

Abstract

Heterogeneous particle swarm optimizers (HPSO) add multiple search behaviors to the swarm. This is done by allowing particles to utilize different update equations to each other. Dynamic and adaptive HPSO algorithms allow the particles to change their behaviors during the search. A number of factors come into play when dealing with the different behaviors, one of which is deciding when a particle should change its behavior. This paper presents a number of behavior changing schedules and strategies for HPSOs. The schedules are compared to each other using existing HPSO algorithms on the CEC 2013 benchmark functions for real-parameter optimization.