Research On Particle Swarm Optimization Algorithm With Strategy Ensemble | Posted on:2016-11-20 | Degree:Master | Type:Thesis | Country:China | Candidate:Z Y Liu | Full Text:PDF | GTID:2298330467492870 | Subject:Operational Research and Cybernetics | Abstract/Summary: | PDF Full Text Request | Since it was proposed, the Particle swarm optimization (PSO) algorithms has attracted much attention from many scholars and got rapid development in some fields related to solving optimization problem, such as machine learning. Most Particle swarm optimization algorithms use a single learning pattern for all particles, which does not benefit to save the heuristic information. A multiple strategies interactive learning mechanism based PSO (DPPSO) is proposed in the third chapter of this paper, which firstly introduces a modified differential mutation strategy to update the velocity of particle in order to diversify the PSO population, then combines differential mutation with the perturbed particle updating strategy as the updating pattern. Each particle generates two intermediate particles from two strategies and selects a better one as its new position. Therefore, each particle can dynamically select its biased generation strategy to diversify the particle population and the searching trajectories. The particles can obtain more beneficial heuristic information, which guides particles to the promising search area.Inspired by the ideas of multi-swarm information sharing and elitist perturbation guiding a novel multi-swarm cooperative multistage perturbation guiding particle swarm optimizer (MCpPSO) is proposed in the fourth chapter of this paper. The multi-swarm information sharing idea is to harmoniously improve the evolving efficiency via information communicating and sharing among different sub-swarms with different evolution mechanisms. It is possible to drive a stagnated sub-swarm to revitalize once again with the beneficial information obtained from other sub-swarms. Multistage elitist perturbation guiding strategy aims to slow down the learning speed and intensity in a certain extent from the global best individual while keeping the elitist learning mechanism. It effectively enlarges the exploration domain and diversifies the flying tracks of particles. The strategy increases the likelihood of jumping out of local optimum, promoting the particles to more accurately locate the global optimal position.Experiments indicate the effectiveness and the reciprocal reinforcement of two strategies. DPPSO has the best comprehensive performance, with powerful global exploration and fine locating abilities, when comparing with other particle swarm optimizers and state-of-the-art algorithms. | Keywords/Search Tags: | Particle swarm optimization, multiple strategies, differential mutation, perturbation strategy, multi-swarm cooperative, information sharing, swarm intelligence, numerical optimization | PDF Full Text Request | Related items |
| |
|