基于八种智能算法的配电网络重构计算平台及其性能  

A Computing Platform for Distribution Network Reconfiguration Based on 8 Intelligent Algorithms and Its Performance

作　　者：高元海[1] 王淳[1] 

机构地区：[1]南昌大学电气与自动化系,江西南昌330031

出　　处：《中国电力》2015年第8期53-60,共8页Electric Power

基　　金：国家自然科学基金资助项目(51167012)~~

摘　　要：采用8种基于群体的智能算法构建了一个求解配电网重构问题的计算平台,以期寻求到一个适合于求解该问题的智能算法。平台中不同算法之间仅算法主体部分不同,配电网重构模型和基本环路搜索等模块完全一致,种群规模、精英数量等公共参数也完全相同。文中给出了各种算法的基本原理和求解步骤,以IEEE 33节点系统为例测试了算法参数的敏感性,对比了各种算法的性能差别,并使用IEEE 16节点和PG&E69节点测试系统对算法的适应性做了进一步比较。测试得到的目标函数平均值、收敛到最优解的比例、计算时间以及对系统规模适应性等方面的结果表明:螺栓遗传算法(Stud GA)性能最优、生物地理学优化算法(BBO)次之、其他算法在不同的测试系统中表现的性能不一致。Stud GA具有操作简单、参数少、收敛到最优解的概率高、计算时间短等优点,适合用于求解配电网重构问题。In order to seek an intelligent algorithm that is suitable for solving the distribution network reconfiguration issue, a computing platform which employs eight population-based intelligent algorithms is constructed. In the platform, for different algorithms, the distribution network reconfiguration model and the basic loop search module are completely consistent, with the parameters such as population size and elite number being identical. The basic principles and calculation procedures of the eight algorithms are given. The sensitivity of the algorithm parameters is tested, and the performance of the eight algorithms is compared using IEEE 33-bus system. In addition, the adaptability of the algorithms is further compared using the IEEE 16-bus and PG＆E 69-bus test systems. The results show that Stud GA is the most suitable algorithm and BBO comes the second in terms of the average objective function value, the probability of converging to the best solution, the computation time and the adaptability to the systems of different scales, and the other algorithms have inconsistent performance in different test systems. Stud GA is a suitable algorithm for solving the distribution network reconfiguration issue because of its advantages of simple operation, few parameters, short computing time, and high probability of converging to the optimal solution.

关 键 词：电力系统 智能算法 配电网重构 计算平台 参数敏感性 螺栓遗传算法 生物地理学优化算法 

分 类 号：TM727[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]