CAJ | 학술논문

在群体智能算法中个体种群的多样性在进化后期逐渐消失，个体趋同性增加，因此粒子群算法的主要缺点是容易陷入局部最优值。提出了一种新的改进粒子群算法，该算法结合了压缩因子和综合信息策略，其中压缩因子可以平衡粒子群算法中的局部和全局搜索，综合信息可以较好地加强种群的多样性。改进后的粒子群算法与基本粒子群算法、自适应粒子群算法和压缩因子粒子群算法在7个测试函数上分别进行了精度对比测试、成功概率测试和收敛速度测试，结果表明新算法获得了较高的搜索精度和较快的收敛速度。
재군체지능산법중개체충군적다양성재진화후기축점소실，개체추동성증가，인차입자군산법적주요결점시용역함입국부최우치。제출료일충신적개진입자군산법，해산법결합료압축인자화종합신식책략，기중압축인자가이평형입자군산법중적국부화전국수색，종합신식가이교호지가강충군적다양성。개진후적입자군산법여기본입자군산법、자괄응입자군산법화압축인자입자군산법재7개측시함수상분별진행료정도대비측시、성공개솔측시화수렴속도측시，결과표명신산법획득료교고적수색정도화교쾌적수렴속도。
The diversity of swarm will be impaired in late period of evolution for a swarm intelligent algorithm and the convergence of each individual element is enhanced, so the major disadvantage of particle swarm optimizer is vulnerable to be trapped in the local optima. This paper proposes a new variant particle swarm optimizer which com-bines constrict factor and comprehensive informed strategy. The constrict factor can balance the global and local models, and comprehensive informed strategy can efficiently enhance the diversity of all particles. By comparing the standard particle swarm optimizer, adaptive particle swarm optimizer and particle swarm optimizer based on con-strict factor on 7 test functions with accuracy level, success rate and convergence velocity, the results show that the new algorithm can obtain a higher accurate level and faster convergence velocity.