Distributed Moving Horizon Estimation via Operator Splitting for Automated Robust Power System State Estimation

In this study, we present methods of optimization-based power system state estimation over sensor networks. By minimizing a composite loss function while ensuring that the state, disturbance, and measurement noise constraints are satisfied, the best or better state estimates are iteratively computed. The proposed distributed computational methods for power system state estimation are based on operator splitting. Our methods are computationally decomposable over sensor networks, so distributed and parallel computing can be applied. They can systematically handle the constraints of the state variables and noise as well as disturbances, such that the negative effects of bad data and parametric model uncertainty can automatically be reduced in the estimation. For demonstration, the IEEE 118-bus power system dynamic state estimation problem is considered. The results are compared to the ones obtained from a distributed extended Kalman filter. It is shown that compared with a distributed extended Kalman filter, the proposed method achieves improved robustness against adversarial data defection.

Automated summary

The optimization-based power system state estimation method proposed in this study, utilizing distributed computing, effectively handles constraints, noise, and disturbances, reducing the negative impacts of bad data and parametric uncertainty.