基于分段电流特征和随机森林算法的低压配电网拓扑识别方法

Topology identification method of low voltage distribution network based on segmented current characteristics and random forest algorithm

低压配电网的拓扑结构对于实现台区线损分析、故障定位、需求响应等应用功能有着重要意义。针对当前低压配网户变关系在线识别难题,提出了一种基于分段电流特征和随机森林算法的低压配电网拓扑识别方法。首先,基于电流特性纵向传导机理,构建电流特征指标体系,并提出基于电流时间序列的分段特征提取方法。其次,根据分段电流特征生成特征向量集,建立基于随机森林算法的拓扑识别模型,可快速有效识别用户连接变压器关系。最后,利用实际配电网的量测数据进行算例分析,结果证明该方法拓扑识别准确率达到98.02%,所需样本时间短,能够快速有效地识别低压配网拓扑关系。

The topology of low-voltage distribution network is of great significance to realize the application functions such as line loss analysis, fault diagnosis, power theft early warning and demand response. Aiming at the problem of inaccurate and missing topology of low-voltage distribution network, a topology identification method of low-voltage distribution network based on segmented current characteristics and random forest algorithm is proposed. Firstly,based on the current similarity mechanism, the current characteristic index is proposed, and the current time series is segmented for feature extraction. Secondly, the feature vector set is generated according to the segmented current characteristics, and the topology recognition model based on random forest algorithm is established,which can quickly and effectively identify the relationship between users and transformers. Finally, an example is analyzed by using the measured data of the actual distribution network. The results show that the topology recognition accuracy of this method is 98.02%, the sample time is short, and it can quickly and effectively identify the topology relationship of low-voltage distribution network.