Serial static–dynamic state estimation for renewable-rich distribution networks

The increasing penetration of intermittent renewable energy sources and diverse loads has significantly increased the operational complexity and uncertainty of modern distribution networks. To address these challenges, this paper proposes a serial static-dynamic state estimation system with adaptive anomaly detection for distribution systems characterized by high renewable energy penetration. In static state estimation (BFS-SE), a dual-state backward/forward sweep-based formulation is developed to jointly estimate node voltages and branch currents, thereby enhancing estimation accuracy and numerical stability. An adaptive bad data detection mechanism is integrated into each backward and forward sweep iteration, effectively reducing false alarms and missed detections under complex measurement conditions. In dynamic state estimation (UKF-SE), an improved unscented kalman filter-based approach is employed with explicit modeling of renewable energy sources. The robust BFS-SE results are used to initialize the UKF-SE, enhancing convergence and tracking stability. Additionally, a Chi-square test-based bad data detection strategy is incorporated into the UKF-SE method to further enhance the robustness of the dynamic estimation process. Quantitative simulation results demonstrate the superiority of the proposed framework: the BFS-SE stage restricts the average voltage magnitude error to 0.33% within just 0.10 s. Subsequently, the UKF-SE stage achieves a highly precise dynamic tracking root-mean-square error (RMSE) of (Formula presented) p.u. Furthermore, the embedded adaptive anomaly detection sustains an outstanding F1-score of 98.46% even under severe cyber-physical stress, confirming the framework’s robust real-time applicability.