Simultaneous State Estimation and Bad Data Detection on PMU Measurements in Active Distribution Power Networks

Abstract

State Estimation calculations and real-time monitoring of distribution power networks has been more difficult due to increased use of Distributed Generations (DGs). Though difficult, tracking dynamic changes is possible in Active Distribution Power Networks (ADNs) by implementing Distribution System State Estimation (DSSE) calculations. DSSE methods rely heavily on real measurements collected by sensing devices which are installed in ADNs. However, the accuracy of real measurements can be diminished in the presence of False Data Injection Attacks (FDIAs). FDIAs can cause inaccurate results of State Estimation (SE), and the management and control of ADNs are commonly influenced by these inaccurate results. To combat these errors, detection methods are applied to identify FDIAs on measurements in ADNs.Existing methods perform state estimation calculations and FDIA detection separately. In this study, we propose a new method to simultaneously perform DSSE calculations and detect FDIAs by applying a novel deep learning approach. For this purpose, regression and classification calculations are performed by a single Deep Neural Network (DNN) model to perform SE calculation and FDIAs detection at once. The results of the proposed method are compared when SE calculation and FDIAs detection are executed separately. The results illustrate that the proposed method can perform state estimation calculation and FDIA detection on the available measurements with high precision. The effectiveness of the proposed method is validated in different criteria by the modified IEEE 33-bus and 69-bus standard distribution system with the consideration of DGs.