基于蒙特卡罗分析的配电网架规划方法比较

Comparison of Expansion Planning Alogithms for Distribution Networks Based on Monte-carlo Simulation

提出了一种基于蒙特卡罗分析的方法,对常用配电网扩展规划方法进行比较。在给定的网格状规划区域上,通过随机设置电源点、负荷和其他参数信息构成各种场景,分别采用各种规划方法进行计算。并采用方差比校验、均值校验和区间估计方法处理所得的结果。定义了对两种方法的优劣、等效和等价进行严格评价的指标。还论述了规划方法的综合比较方法。对基本最小生成树法、改进最小生成树法、支路交换法和遗传算法等方法进行了比较,结果表明:改进最小生成树法在规划效果上与遗传算法等效,但是计算效率显著优于遗传算法。支路交换法的规划效果不如遗传算法和改进最小生成树算法,其规划效率显著优于遗传算法但仍比改进最小生成树算法差。改进最小生成树法的综合效果最佳。实例表明提出的方法是可行的。

Comparisons of several algorithms for expansion planning of distribution networks are made based on a Monte-Carlo simulation based approach. Various scenarios are formed on a given grid by randomly setting main position, loads and other parameters. Planning result of each algorithm is obtained on each scenario, respectively. The variance ratio test, mean value test and interval estimations are introduced in analysis of the results. A group of indices are defined in strict evaluation between each two algorithms, such as Superior, Inferior, Equivalent and Equal. A hybrid approach of comparison is also described. Comparisons of basic, improved minimum-cost spanning tree based algorithms, Branch Exchange algorithm （BE） and Genetic Algorithm （GA） are carried out. The comparison results show that Improved Minimum-cost Spanning Tree based algorithms （IMST） are not only equivalent to GA in planning effects but also superior to GA in efficiency. The planning result of BE is worse than IMST and GA. The efficiency of BE is between that of IMST and GA. The hybrid features IMST based algorithms are the best. Above examples show the feasibility of the proposed methods.