Accelerating particle swarm optimization using crisscross search
| Autor: | Anbo Meng, Zhuangzhi Guo, Sizhe Chen, Zhuan Li, Hao Yin |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
education.field_of_study
Mathematical optimization Information Systems and Management Optimization problem Computer science 020209 energy Crossover Population MathematicsofComputing_NUMERICALANALYSIS Particle swarm optimization Swarm behaviour 02 engineering and technology ComputingMethodologies_ARTIFICIALINTELLIGENCE Computer Science Applications Theoretical Computer Science Maxima and minima Local optimum Artificial Intelligence Control and Systems Engineering 0202 electrical engineering electronic engineering information engineering Multi-swarm optimization education Software |
| Zdroj: | Information Sciences. 329:52-72 |
| ISSN: | 0020-0255 |
| DOI: | 10.1016/j.ins.2015.08.018 |
| Popis: | This paper introduces a novel crisscross search particle swarm optimizer called CSPSO.The CSPSO algorithm has significant superiority over most of the other PSO variants in terms of solution accuracy and convergence rate.The swarm in CSPSO is directly represented by a population of pbests, which are renewed by the modified PSO search as well as the crisscross search in sequence at each generation.The CSO as an catalytic agent can accelerate the particles to converge to the global optima.The horizontal crossover uses a cross-border search mechanism to enhance the global search ability greatly.The vertical crossover can facilitate the stagnant dimensions to escape out of the local minima. Although the particle swarm optimization (PSO) algorithm has been widely used to solve many real world problems, it is likely to suffer entrapment in local optima when addressing multimodal optimization problems. In this paper, we report a novel hybrid optimization algorithm called crisscross search particle swarm optimization (CSPSO), which is different from PSO and its variants in that each particle is directly represented by pbest. The population of particles in CSPSO is updated by modified PSO as well as crisscross search optimization (CSO) in sequence within each iteration. CSO is incorporated as an evolutionary catalytic agent that has powerful capability of searching for pbests of high quality, therefore accelerating the global convergence of PSO. CSO enhances PSO by two search operators, namely horizontal crossover and vertical crossover. The horizontal crossover further enhances PSO's global convergence ability while the vertical crossover can enhance swarm diversity. Several benchmark functions are used to test the feasibility and efficiency of the proposed algorithm. The experimental results show that CSPSO outperforms other state-of-the-art PSO variants significantly in terms of global search ability and convergence speed on almost all of the benchmark functions tested. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect