Simple and Adaptive Particle Swarms

摘要: The substantial advances that have been made to both the theoretical and practical aspects of particle swarm optimization over past 10 years taken it far beyond its original intent as a biological swarm simulation. This thesis details explains these in context what has been achieved this point, well yet be understood or solidified within research community. Taking into account state modern field, standardized PSO algorithm is defined for benchmarking comparative purposes work, for community whole. This standard refined simplified several iterations form does away with potentially undesirable properties while retaining equivalent superior performance on common set benchmarks. refinement, referred discrete recombinant swarm (PSODRS) requires only single user-defined parameter positional update equation, uses minimal additive stochasticity, rather than multiplicative stochasticity inherent PSO. After a mathematical analysis PSO-DRS algorithm, an adaptive framework developed rigorously tested, demonstrating effects tunable particle- swarm-level parameters. adaptability shows benefit by broadening range problems which wellsuited to optimize.