CAJ | 학술논문

针对粒子群算法的“早熟”，进化后期收敛速度慢及精度低等问题，提出了一种改进的PSO算法。为保证初始群体的遍历性，改进算法首先利用了信息熵产生初始群体；为提高进化过程中群体的多样性，将遗传算法中杂交、变异的思想融入了算法中；为提高算法晚期的收敛速度，将模拟退火算法中退火的思想引入到杂交过程中。该算法与其他改进算法进行数值比较，仿真实验表明，提出的算法抗“早熟”能力强，搜索精度高，稳定性好。
침대입자군산법적“조숙”，진화후기수렴속도만급정도저등문제，제출료일충개진적PSO산법。위보증초시군체적편력성，개진산법수선이용료신식적산생초시군체；위제고진화과정중군체적다양성，장유전산법중잡교、변이적사상융입료산법중；위제고산법만기적수렴속도，장모의퇴화산법중퇴화적사상인입도잡교과정중。해산법여기타개진산법진행수치비교，방진실험표명，제출적산법항“조숙”능력강，수색정도고，은정성호。
For the disadvantages of Particle Swarm Optimization(PSO)algorithm, such as“premature”, slow convergence speed and low convergence precision in the late evolutionary, a novel hybrid PSO algorithm is proposed. The algorithm applies the entropy theory to generate the initial population, and integrates simulated annealing and the ideas of hybridiza-tion and mutation in the genetic algorithm into evolutionary process. Results of simulation experiment show that the algo-rithm compared with other modified PSO algorithms proposed before is improved virtually on anti“premature”ability, optimization precision and stability.