基于量测数据质量的低压台区拓扑识别结果可信度评价

在没有先验拓扑连接关系的情况下,基于数据分析法的低压拓扑识别准确率无法量化评估.为此从数据分析法所需的量测数据质量着手,提出了一种根据高级量测体系(AMI)量测质量评价出数据分析法识别结果可信度的方法.首先,通过剖析基于二次规划的数据分析法的工作原理,推导数据质量对识别准确率的影响机理.其次,从数据完整性、数据差异性和数据时间尺度3个方面出发,建立与识别准确率关联的数据质量评价模型,量化了所需数据的质量.然后,以数据质量评价分数和相应的拓扑识别准确率为训练样本,搭建可根据AMI量测质量评价出数据分析法识别准确率范围的模型.最后,以中国广东省3个典型低压台区的数据构建测试集,验证了所提模型的有效性和实用性.

Credibility Evaluation of Topology Identification Results in Low-voltage Distribution Network Based on Quality of Measured Data

Without a priori topologic connection relationship, the accuracy rate of the low-voltage topology identification based on the data analysis method is difficult to evaluate quantitatively. Therefore, starting from the measured data quality required by the data analysis method, a method that can evaluate the credibility of identification results of the data analysis method based on the data quality measured by advanced metering infrastrucuture (AMI) is proposed. Firstly, the working principle of the data analysis method based on quadratic programming is analyzed, and the influence mechanism of the data quality on the identification accuracy is deduced. Secondly, considering three aspects of data integrity, data difference and data time scale, a data quality evaluation model highly correlated with identification accuracy is calculated, which quantifies the quality of the required data. Then, regarding the data quality evaluation score and the corresponding topology identification accuracy as the training sample, a model is built, which can judge the accuracy evaluation of the data analysis method based on the data quality measured by AMI. Finally, the effectiveness and practicability of the proposed model are verified by a test set constructed with the data of three typical low-voltage distribution networks in Guangdong Province of China.