Power consumption abnormity diagnosis method and system based on thickness clustering and big data

The invention belongs to the technical field of smart power grids, and particularly relates to a power consumption abnormity diagnosis method and system based on thickness clustering and big data. Performing thickness clustering division on the acquired line data to obtain different types of power distribution network line data; extracting data characteristics of the obtained different types of power distribution network line data; according to the extracted data features and a preset power consumption diagnosis model, the obtained power distribution network line data are evaluated, and diagnosis of power consumption abnormity is achieved; wherein the preset power consumption diagnosis model adopts an optimized high-order map attention network based on a residual convolutional neural network. 本发明属于智能电网技术领域，具体涉及一种基于粗细度聚类和大数据的用电异常诊断方法及系统，包括：获取配电网线路数据；对所获取的线路数据进行粗细度聚类划分，得到不同类别的配电网线路数据；提取所得到的不同类别配电网线路数据的数据特征；根据所提取的数据特征和预设的用电诊断模型，评估所获取的配电网线路数据，实现用电异常的诊断；其中，预设的用电诊断模型采用基于残差卷积神经网络的优化高阶图注意力网络。