MULTI-MODAL, MULTI-DISCIPLINARY FEATURE DISCOVERY TO DETECT CYBER THREATS IN ELECTRIC POWER GRID

摘要

According to some embodiments, a plurality of heterogeneous data source nodesmay each generate a series of data source node values over time associatedwith operationof an electric power grid control system. An offline abnormal state detectionmodelcreation computer may receive the series of data source node values andperform a featureextraction process to generate an initial set of feature vectors. The modelcreation computermay then perform feature selection with a multi-model, multi-disciplinaryframework togenerate a selected feature vector subset. According to some embodiments,featuredimensionality reduction may also be performed to generate the selectedfeature subset. Atleast one decision boundary may be automatically calculated and output for anabnormalstate detection model based on the selected feature vector subset.