Data-Driven Bayesian-DMD Based Method for Regional Inertia Assessment in Power Systems

Accurate inertia estimation is essential for the operational stability and control of power systems, particularly under low-inertia conditions. A precise assessment of system inertia supports reliable grid management and enhances system resilience against disturbances. Existing evaluation methods mainly rely on offline data or specific disturbance events, which are sensitive to noise and lack real-time. In this paper, a regional inertia evaluation method based on Bayesian dynamic mode decomposition (Bayesian-DMD) is proposed. By extending the dynamic modal decomposition to the Bayesian probability framework, combined with Gibbs sampling technology, a posterior distribution estimation model of modal parameters of noisy data is constructed to handle the measured noise and model uncertainty effectively. Based on the reconstructed equivalent second-order sway equation, the inertia parameter extraction mechanism is established to improve the robustness of the noise scene. Finally, the effectiveness and robustness of the proposed algorithm are verified by using the IEEE 4-machine 2-zone system example.