一类基于Multi-Agent的实时进化算法  

A Real-time Evolutionary Algorithm Based on the Multi-Agent Theory

作　　者：王健 

机构地区：[1]大庆华科股份有限公司聚丙烯分公司,黑龙江大庆163316

出　　处：《机电产品开发与创新》2014年第1期15-17,共3页Development & Innovation of Machinery & Electrical Products

摘　　要：实时进化(Real-Time Evolutionary,RTE)策略解决了传统实时优化(Real-Time Optimization,RTO)方案中等待稳态的缺点,受RTE思想的启发,论文提出了一种基于Multi-Agent的实时进化算法。首先将粒子群算法与Multi-Agent机制相结合,每一个Agent相当于粒子群算法中的一个粒子,通过和其邻居进行竞争、合作以及学习,能够迅速、准确的找到全局最优解;然后,根据RTE思想,将基于Multi-Agent的粒子群算法应用于RTO的解决方案。通过对Williams-otto反应器的实例研究,证明了所提算法的有效性。The Real-Time Evolutionary（RTE）can be utilized to address the demerit of the traditional Real-Time Optimization （RTO） which is the requirement for waiting the steady. Based on this conception, this paper proposed a real-time evolutionary algorithm regarding the Multi-Agent theory. Firstly, a combination between the Particle Swarm Optimization（PSO） algorithm and the Multi-Agent theory is demon- strated, in which each Agent can be treated as a particle within the PSO, which can quickly and accurately find the global optimal solution through competition, cooperation, and learning with their neighbors. Then, derived from the conception of RTE, the Multi-Agent based PSO algorithm is applied on the RTO problem. In the end, the effectiveness and availability of this algorithm are confirmed by the empirical test of Williams-otto reactor.

关 键 词：MULTI—AGENT RTE RTO 

分 类 号：TP317[自动化与计算机技术—计算机软件与理论]