Topology-Linked Self-Adaptive Quantum Particle Swarm optimization for Dynamic Environments
Rethabile Mabaso, Christopher W. Cleghorn · 2020
The canonical particle swarm optimizer (PSO) is generally unsuitable for use in dynamic environments owing to its loss of diversity, the inability to detect changes in the environment, and the outdated memory of the particles' previous positions following such a change. In an attempt to overcome these challenges, the quantum particle swarm optimizer (QPSO) was developed. However, QPSO introduces an additional problem-dependent parameter called the “cloud radius” that requires problem-specific tuning. A recent variation of the QPSO that does not use the cloud radius as a parameter has been demonstrated to match and at times surpass the performance of QPSO. This variation has however only been tested using the ring topology, despite the self-adaptive mechanism utilized having a strong coupling to the used topology. This study aims to investigate the effects of various network topologies on the performance of self-adaptive QPSO. A topology based on the 3D von Neumann structure has been shown to perform significantly better than other topologies.