原始粒子 meaning in Chinese
direct particle
primary particle
progenitor
Examples
- These are loose aggregates of a large number of primary particles of crystalline or spherical shape .
这些是大量晶体，或球原始粒子的疏松团粒。 - But , pso convergence ' s speed become slow in latter iterative phase , and pso is easy to fall into local optimization . at present , some scholars improve base pso mostly using 3 methods : disperse algorithm , increase convergence speed , enhance particle ' kinds . in the paper , i put forward 2 methods aiming at local best resutl but not whole best result . i modify base pso using the last method . some scholars put forward times initializations , so i select best result after circulating some times to be a parameter of formula . first , put particle into some small region , and ensure every region having one paticle at least . second , every region ' s particle has probability transfer other regions . although increase running time , enhance particle ' kinds , decrese the probability of convergence far from whole best result . nerms ( network educational resource management system ) is one of the research projects in the science and technology development planning of jilin province . the aim of nerms is to organize and manage various twelve kinds of network educational resources effectively so that people can share and gain them easily and efficiently , so as to quicken the development of network education
但粒子群算法仍存在如下不足:首先在多峰的情况下,粒子群有可能错过全局最优解,远离最优解的空间,最终得到局部最优解;其次在算法收敛的情况下,由于所有的粒子都向最优解的方向群游,所有的粒子趋向同一,失去了粒子间解的多样性,使得后期的收敛速度明显变慢,同时算法收敛到一定精度时,算法无法继续优化,本文对原始粒子群算法提出了二点改进方案: 1 .算法迭代到一定代数后,把此时找到的全局最优解当作速度更新公式的另一参数(本文称之为阶段最优解)再进行迭代; 2 .每次迭代过程中除最优解以外的每个粒子都有一定概率“变异”到一个步长以外的区域,其中“变异”的粒子在每一维上都随机生成一个步长。