Distribution System State Estimation and False Data Injection Attack Detection with a Multi-Output Deep Neural Network

Distribution system state estimation (DSSE) has been introduced to monitor distribution grids; however, due the incorporation of distributed generations (DGs), traditional DSSE methods are not able reveal operational conditions active networks (ADNs). calculation depends heavily on real measurements from measurement devices in networks. However, accuracy and results can be significantly affected by false data injection attacks (FDIAs). Conventional FDIA detection techniques often unable identify FDIAs into data. In this study, a novel deep neural network approach is proposed simultaneously perform (i.e., regression) binary classification) using measurements. work, classification nodes DNN allow us which phasor unit (PMU), if any, were affected. approach, we aim show that method available with high accuracy. We compare our detecting performing SE calculations separately; moreover, compared weighted least square (WLS) algorithm, common model-based method. The achieves better performance than WLS separate DSSE/FDIA presence erroneous measurements; also executes faster other methods. effectiveness validated two schemes case studies: one modified IEEE 33-bus without DGs, 69-bus DGs. illustrated F1-score when only. successfully detected each PMU measurement. Moreover, regression-only method, bad