摘要
针对常规遗传算法 (SGA)的不足 ,采用新颖的智能体 (Agent)技术构建多Agent系统实现遗传算法 ,能从进化环境中获取表征当前进化状态的有用信息 ,智能地监控调度GA的进化操作 ,在避免早熟的同时加快全局寻优 ,提升GA的优化性能 ,对于复杂问题其优势更为显著 .开发了基于多Agent的遗传算法 (M Agent GA) ,并应用于二甲苯异构化装置的操作优化 。
Genetic algorithms (GA) are used more and more wide in the optimization of complicated, nonlinear questions in the chemical industries. A multi-Agent system which includes two kinds of Agents: O-Agent and A-Agent is built. There are a number of O-Agents in the system, and their function is equal to the seeds of evolution, but there is only one A-Agent. The A-Agent is in charge of the work of evolution. It monitors and controls the GA evolution procedure. These two kinds of Agents can exchange the useful information which can represent the current situation of evolution. By means of the limit of life of these two kinds of Agents, the system can avoid prematureness and at the same time can speed up convergence. The genetic algorithms based on Intelligent Agent (M-Agent-GA) improves greatly the feature of standard GA (SGA), especially in the optimization of complicated questions. It was used in the optimization of the equipment of xylene isomerization and the result was good.
出处
《化工学报》
EI
CAS
CSCD
北大核心
2003年第5期653-658,共6页
CIESC Journal
基金
国家自然科学基金资助项目 (No 2 0 0 760 41)~~