摘要： 在研究粒子群优化算法生物特征的基础上,提出了粒子群优化算法的异步模式。在异步模式的程序实现上,采用Java多线程技术,使每个粒子的行为成为一个独立的线程,进化中的粒子个体充分表现出独立性,种群表现出异步性。最后利用一些经典的标准测试函数,与经典PSO算法(可称为同步模式)进行了比较分析,结果表明:异步模式的收敛速度较同步模式有显著的提高;同时,在一个较小时间段(一 般小于整个运算时间的5%)之后,异步模式在寻优效果上也明显优于同步模式。

关键词: 粒子群优化算法, 异步, Java, 多线程

Abstract: This paper proposes an asynchronous pattern from analyzing on the biologic character of particle swarm optimization. The asynchronous pattern is programmed with Java multiple threads, and considered each of the particles as a thread. As a result, each particle acts as an independent individual and the searching of population performs asynchronously. The experiments of some standard testing functions show that it is obviously enhance on the operation efficiency by this asynchronous pattern.

中图分类号: 

• TP311.1