Feature identification method for abnormal operations in grid dispatch operations based on association analysis

The premise of identifying abnormal operations in power grid dispatching operations is to identity features of abnormal dispatch operations. This paper proposes a feature recognition method for abnormal operations in dispatching operations based on association analysis, laying the foundation for identifying abnormal operations. Firstly, historical operational data is preprocessed. Considering the scarcity of abnormal operations in historical data, data augmentation is performed to increase the proportion of abnormal operations. Secondly, in order to eliminate the influence of redundant information such as place names and equipment numbers in the data, clustering algorithm is used to fuzzify the operation content and operation time in the operational data. Thirdly, association analysis algorithm is used to extract compliant/abnormal features from data, and the identification of abnormal operations is accomplished using these features. The paper identifies abnormal features from the perspective of data analysis. In the case studies, we conducted feature recognition on the historical operational data of a certain dispatch automation main station in China. The obtained abnormal features include early morning, deputy director of dispatching, senior professional title, system maintenance personnel and operation of knife switch.