Parallel particle swarm optimization algoritms in nuclear problems
Year: 2009
Type: article
Abstract: Particle Swarm Optimization (PSO) is a population-based metaheuristic (PBM), in which solution candidates \nevolve through simulation of a simplified social adaptation model. Putting together robustness, efficiency and \nsimplicity, PSO has gained great popularity. Many successful applications of PSO are reported, in which PSO \ndemonstrated to have advantages over other well-established PBM. However, computational costs are still a \ngreat constraint for PSO, as well as for all other PBMs, especially in optimization problems with time \nconsuming objective functions. To overcome such difficulty, parallel computation has been used. The default \nadvantage of parallel PSO (PPSO) is the reduction of computational time. Master-slave approaches, exploring \nthis characteristic are the most investigated. However, much more should be expected. It is known that PSO \nmay be improved by more elaborated neighborhood topologies. Hence, in this work, we develop several \ndifferent PPSO algorithms exploring the advantages of enhanced neighborhood topologies implemented by \ncommunication strategies in multiprocessor architectures. The proposed PPSOs have been applied to two \ncomplex and time consuming nuclear engineering problems: i) reactor core design (CD) and ii) fuel reload (FR) \noptimization. After exhaustive experiments, it has been concluded that: i) PPSO still improves solutions after \nmany thousands of iterations, making prohibitive the efficient use of serial (non-parallel) PSO in such kind of \nrealworld problems and ii) PPSO with more elaborated communication strategies demonstrated to be more \nefficient and robust than the master-slave model. Advantages and peculiarities of each model are carefully \ndiscussed in this work.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2298259999