基于双重相似性与自适应风险权重配置的配电网运行风险识别

Risk Identification of Distribution Network Operation Based on Dual Similarity and Adaptive Risk Weighting Configuration

配电网的运行风险识别对应急供电、电力系统运行维护、扩容等具有重要指导意义。由于配电网运行条件复杂、数据测量不充分、数据无标记和数据不平衡等问题,给传统的基于逻辑判断的配电网风险识别方法带来了挑战。文章从线路侧跳闸预警、营销侧宏观预测两方面着手,提出了一种综合识别配电网运行态势的风险识别框架。首先设计了一种基于双重相似性机制的配电跳闸风险识别方法,通过基于K-means的矩阵剖面(matrix profile)重构运行模式并衡量运行模式的相似性,基于时序高分辨率连续测量的优势进行风险识别;同时,提出了一种基于权重自适应配置的配电营销宏观预测算法。以中国某南方城市地区配电网的生产数据为基础,对各种风险事件进行了预测与识别,验证了算法的效果优于传统的配电网风险识别方法。

The identification of operational risks in distribution network has important guiding significance for emergency power supply,power system operation and maintenance,and expansion.Due to the complex operating conditions,insufficient data measurement,unmarked data,and imbalanced data in distribution network,traditional logic-based risk identification methods for distribution network face challenges.This paper proposes a risk identification framework for comprehensive operational situation of distribution network from two aspects:line side trip warning and marketing side macro prediction.Firstly,a distribution tripping risk identification method based on dual similarity mechanism is designed.The operating mode is reconstructed using a matrix profile based on K-means and the similarity of the operating modes is measured.Based on the advantage of high-resolution continuous measurement in time series,risk identification is carried out.Meanwhile,a macro prediction algorithm for distribution marketing based on weight adaptive configuration is proposed.Based on production data of a distribution network in a southern urban area of China,this paper predicts and identifies various risk events,verifying that the algorithm is more effective than traditional distribution network risk identification methods.