International Journal of Computers Communications & Control · 2012 · 41 citations · 18 references
EngineeringComputational EngineeringIndustrial EngineeringOptimizationProcess ControlProcess Systems EngineeringController TuningEvolutionary ComputationMetaheuristic OptimizationMutation OperatorsOptimal ControlPopulation-based SearchConvergenceEvolutionary AlgorithmsExplorationHybrid Evolutionary AlgorithmsParticle Swarm ControlParticle Swarm OptimizationPid TuningPremature Convergence
Particle swarm optimization (PSO) has proved its ability as an efficient search tool in many optimization problems. However, PSO is easy to be trapped into local minima due to its mechanism in information sharing. Under this circumstance, all the particles could quickly converge to a position by the attraction of the best particle; all particles could hardly be improved. To overcome premature convergence of the standard PSO algorithm, this paper presents an adaptive hybrid PSO, namely (AHPSO) by employing an adaptive mutation operator for local best particles instead of applying the mutation operator to the global best particle as has been done in previous work. The developed algorithm is a new approach which allows the swarm to be more diverse by making better exploration of the local search space instead of global search space investigated by previous researchers. The proposed algorithm holds on the properties of simple structure, fast convergence, and at the same time enhances the variety of the population, and extends the search space. It is applied to self-tuning of proportional-integral-derivative-(PID) controller in the ball and hoop system which represents a system of complex industrial processes. The results are compared with those obtained by applying standard PSO, and adaptive hybrid PSO based on global best particles. It has been shown that the developed AHPSO local best algorithm is faster in convergence and the obtained results are proved to have higher fitness than the other two algorithms.
18
Riccardo Poli, James Kennedy, Tim Blackwell · Swarm Intelligence · 2007 · 21.4K citations
Karl Johan Åström, Tore Hägglund · Control Engineering Practice · 2001 · 1.4K citations
Using selection to improve particle swarm optimization
Peter J. Angeline · 2002 · 872 citations