基于最小生成树及改进遗传算法的含分布式电源配电网孤岛划分方法

Islanding Method Based on Minimum Spanning Tree and Improved Genetic Algorithm for Distribution System with DGs

为在配电网大规模失电条件下充分发挥分布式电源作用,减小停电损失,基于分布式电源孤岛在线划分的思想,提出了配电网故障后分布式电源孤岛划分的原则,建立了分布式电源孤岛划分的数学模型。提出了孤岛划分算法中无向图的权值整定方法,在此基础上提出分为2个步骤的孤岛划分方法:第1步为基于Kruskal算法的含分布式电源最优辐射状配电网的生成方法;第2步为基于改进遗传算法的多DG配电网最优孤岛划分方法。基于PG&E69节点系统对负荷权重设置对算法的影响进行了分析,并对各种方法进行了对比验证。结果表明,最优辐射状配电网是进行最优孤岛划分的基础,可缩小解集空间;所提方法相比启发式算法、分支定界法和动态规划法在恢复负荷数量、负荷价值方面具有优越性,可应用于配电网运行控制。

In order to make full use of distributed generators and reduce power loss under the condition of large scale power loss in distribution network, on the basis of idea of on-line islanding algorithm, we proposed the principles of is- landing algorithm and introduced the mathematic model. We proposed a weight value setting method for the islanding algorithm. An optimal radial distribution network generation method based on Kruskal algorithm produces the optimal radial distribution network, by which the best islanding scheme is proposed by improved genetic algorithm. PG＆E 69 nodes system is used to analyze the influence of load weight setting and compare the existing methods. The result shows that optimal radial distribution network can reduce the solution space, which is the basis of optimal islanding algorithm. Compared with heuristic, branch and bound and dynamic programming method, the proposed two-step method is more effective in quantity and value of the loads restored, which can be used in distribution network operation control.