摘要
提出一种配电网动态重构的新方法,并利用多代理系统(multi-agentsystem,MAS)技术来实现。将一天分为多个时段,每一时段由一个工作代理负责,利用混合粒子群算法进行静态重构,求得一个解集,并通过相邻代理间的相互学习,对解集进行补充和完善。然后对解集进行评估,从中选出种子解和优先替补解,计算出每一工作代理减少一次开关操作将增加的最小损耗。最后经协调代理多次迭代协调,令部分工作代理更换种子解,直到满足开关操作次数约束条件为止,此时,各工作代理的种子解即共同组成了动态重构结果。给出MAS的系统框架和各代理的功能模块。最后对一个IEEE69节点标准算例和一个实际算例分别进行仿真,仿真结果表明该方法不仅可快速获得不同约束条件下的最优解,还能为相关部门制定合理的开关操作次数提供依据。
A novel algorithm for the dynamic reconfiguration of the distribution network was proposed based on multi-agent system in this paper. One day was divided into several time intervals and each was managed by a work agent. A hybrid particle swarm optimization algorithm was applied in each time interval to carry on the static reconfiguration and obtain a solution set. The neighboring agents mutually studied from each other to enrich and improve their solution sets. Then each work agent evaluated its solution set, chose the seeded solution and the priority substitute solution from the set, and calculated the minimum loss increase caused by reducing one switch operation. Finally some work agents were selected by the coordination agent to change their seed solutions in order to satisfy the switching operation times constraint and then the seed solution of every work agent constituted the dynamic solution. The simulation results show the effectiveness of the proposed dynamic reconfiguration method.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2008年第34期72-79,共8页
Proceedings of the CSEE
基金
高等学校博士学科点专项科研基金项目(20060294019)。
关键词
配电网
动态重构
多代理系统
混合粒子群算法
distribution network
dynamic reconfiguration
multi-agent system
hybrid particle swarm optimization