基于GRA-K-means++算法的配电网供电可靠性水平评价

Evaluation of Distribution Network Power Supply Reliability Level Based on GRA-K-means++Algorithm

为有效衡量地区电网建设管理水平,差异化地对地区可靠性水平进行量化评价,提出了一种基于GRA-Kmeans++算法的配电网供电可靠性水平评价方法。首先,综合考虑外部因素与电网因素,构建地区配电网可靠性影响因素集。结合地区供电范围,基于灰色关联算法(grey relation analysis,GRA)分析可靠性影响因素与供电可靠性的关联度。然后,提出基于外部因素的地区加权聚类模型,选取外部因素的信息熵和关联度作为指标综合权重,结合带权重的K-means++算法对同等外部条件地区进行聚类,提出可靠性偏移度作为地区可靠性水平评价指标。随后结合电网因素的关联度评价指标,辨识影响地区可靠性偏低的关键因素,确定提升电网可靠性的建设管理方向。采用实际配电网数据进行算法验证,结果表明所提方法能够有效实现配电网可靠性水平评价,并为决策可靠性提升手段提供一定的理论依据。

In order to effectively measure the management level of regional power grid construction and quantitatively evaluate the regional reliability level differentiatedly,a method for evaluating the reliability level of distribution network power supply based on GRA-K-means++algorithm is proposed.Firstly,both external factors and power grid factors are taken into account,the set of reli⁃ability influencing factors of regional distribution network is constructed.The correlation between reliability influencing factors and power supply reliability is analyzed based on grey relational analysis(GRA),combined with the regional power supply range.Next,the regional weighted clustering model is proposed based on the external factors,the information entropy and correlation with reli⁃ability of external factors are used as the index comprehensive weight,and the weighted K-means++algorithm is combined to cluster the regions with the same external conditions.The reliability deviation index is proposed as the evaluation index of regional reli⁃ability level.Then,based on the reliability correlation degree of power grid factors,the key factors leading to low regional reliability are identified,and the direction of construction and management for improving the reliability of power grid is put forward.The algorithm is verified by actual distribution network data,and the results show that the proposed method can effectively realize the reli⁃ability level evaluation of distribution network,and provide a certain theoretical basis for decision about reliability improvement methods.