Transformer unsupervised operation state identification method based on clustering feature tree

The invention discloses a transformer unsupervised operation state identification method based on a clustering feature tree. The transformer unsupervised operation state identification method is usedfor representing the real-time operation state of a transformer. According to the method, clustering feature tree modeling is carried out according to online monitoring data of the transformer, and astate mode and features of the transformer during stable operation are explored; The online monitoring data at the current moment is analyzed and calculated. The state mode and the characteristics arematched and an identification result of the running state is given. The method can assist the short-term operation and online scheduling decision of the power grid, further improves the lean management level of power enterprise equipment, and provides data support for an overhaul operation and maintenance mode. 本发明公开了一种基于聚类特征树的变压器无监督运行状态识别方法，用于表征变压器的实时运行状态。该方法是根据变压器在线监测数据进行聚类特征树建模，发掘变压器稳定运行时的状态模式和特征；然后对当前时刻在线监测数据进行分析计算；最后匹