A New PSO Scheduling Simulation Algorithm Based on an Intelligent Compensation Particle Position Rounding off

Wen-bin Hu,Jia-xing Song,Wen-jie Li

DOI: https://doi.org/10.1109/ICNC.2008.605

2008-01-01

Abstract:The PSO algorithm belongs to the consecutive space optimizing family, whereas, a scheduling problem is a typical discrete space, non-numeral optimizing problem. What kind of particle representing method should be used to map the solution of a scheduling problem; how to map between consecutive space where the PSO falls and discrete space where the solution of a scheduling problem falls; how to design and improve the PSO algorithm; how to adjust the PSO algorithm's parameters to make it work for a scheduling problem; how on earth the PSO algorithm will behave on the scheduling problems, still need to be investigated. Therefore in this paper, in accordance with the characteristics of the scheduling problems, we put forward an appropriate scheme to generate the schedule sequence indirectly by decoding the particles, and we also proposed a new particle representing method called Intelligent Compensation Particle Position Rounding off (ICPPR). Each particle corresponds to an Agent, and the population of particles forms a particle coalition, so a multi-agent coalition forms meanwhile. Therefore, the intelligent compensation rounding-off operations for each particle in the coalition is actually a negotiation between multi-agent coalitions. Finally, the PSO algorithm based on the ICPPR particle representing method had been used for a river scheduling problem, the calculation results showed that Multi-Agent particle swarm algorithm based on the ICPPR has the obvious advantages in the algorithm calculation cost and stability.