Collaborative Line-Transformer Relationship Identification Based on Shapelet Transformation in IoT

The correct identification of the line-transformer relationship in distribution networks is the basis for the management of line loss, operations, fault location, and maintenance. Previous methods are difficult to track topology changes in real time based on historical voltage data, and the recognition accuracy needs to be further improved. With the construction of the power Internet of Things (IoT), efficient and accurate identification is possible. Combined Power IoT terminal collaboration technology, a classification method based on shapelet transform is proposed to accurately identify the line-transformer relationship in real time. First, since there will always be some missing voltage data from the transformer terminal, we propose Gaussian fitting imputation. Then, due to the large amplitude difference in the voltage data collected on site, the transformer terminals with known relationships are used to collaborate, and the voltage characteristic matrix is calculated in parallel based on the shapelet transform algorithm. Furthermore, the matrix is transmitted to the unknown relationship transformer terminal using support vector machine(SVM) for relationship identification. With the proposed method, simulation and on-site experiments results show that the shapelet transform is effective for line-transformer identification on the basis of terminal collaboration.