基于差分Lasso的低压台区拓扑识别方法

Low-voltage transformer topology identification method based on differential Lasso

用电信息采集系统中台区档案的正确性至关重要,人工梳理的方式耗时耗力,而依靠设备的方法成本较高。为提供高效便捷且经济的台区档案识别技术,提出一种基于差分Lasso的低压拓扑识别方法,仅使用台区的用电信息即可完成自动拓扑识别。首先,对用电数据建立Lasso的统计模型,根据表箱内电表属于同一台区的特点,进一步构建差分Lasso模型。然后,通过轮询的方式依次对每个台区计算基于差分Lasso的回归系数,并根据阈值挑选表计。应用该文方法对现场采集的用电数据进行分析,其拓扑识别的结果说明了该文方法的有效性。

It is crucial to the correctness of transformer archives in power consumption information acquisition system.The manual combing mode is time-consuming and labor-consuming,while the method of relying on equipment leads to a high cost.To provide efficient,convenient and economic technology for transformer archives identification,a low-voltage topology identification method based on differential Lasso(Diff-Lasso)is proposed in this paper,with which the automatic topology identification can be implemented only by using the power consumption information.First,the Lasso statistics model is built for the power utilization data and the Diff-Lasso model is further constructed according to the characteristic that the ammeters in one meter box belong to the same transformer.Then,the regression coefficient based on Diff-Lasso is calculated in a polling form in turn for each transformer,and the meters can be selected according to a certain threshold.The proposed method is used to analyze the practical power consumption data,and the results of topology identification show the effectiveness of this method.