Bad-Data-Resilient Dynamic State Estimation for Power Systems with Partially Known Models

State estimation plays a crucial role in smart grids as it provides valuable insights into the grid’s operating state. This paper proposes dynamic state estimation techniques to detect measurement outliers and cyber attacks against phasor measurement units. Specifically, we formulate a recursive optimal state estimator for power systems modeled by nonlinear, underdetermined differential algebraic systems and suggest a residual-based scheme to detect anomalous data. Unlike for static state estimation, the analysis shows that if there are enough available dynamic models, a static stealth attack does not exist. This additional detection layer is achieved by exploiting the inconsistencies between the available dynamic models and online measurements. The proposed methodology is validated on the IEEE 39-bus test system showing the effectiveness of the approach in detecting measurement outliers and stealth attacks. The results highlight the importance of incorporating dynamic models into the power system state estimation to enhance its cyber security.