Abstract:

The volatility and rapid variation of renewable generation pose new challenges to frequency control in new-type power systems. To address this problem, this paper proposes a novel regional grid frequency-control method that combines model predictive control with stochastic optimization techniques. First, a prediction-driven frequency-control framework is established for the regional grid, in which the tie-line power deviations caused by neighboring-area frequency fluctuations are treated as local disturbance variables, effectively decoupling the local frequency-control process from neighboring regions. Next, an online estimation scheme for system inertia constant is developed via system-parameter identification. Together with an adaptive online kernel density estimation technique, probabilistic prediction models are constructed for unit inertia time constants, load demand, renewable power output, and neighboring-area frequency deviations, enabling accurate representation of multiple uncertainties. Building on these models, a stochastic-model-predictive-control-based frequency control optimization model and its fast solution algorithm are formulated. Finally, the proposed method is validated on a modified IEEE-39-bus system. The case study results show that, compared with the deterministic strategy and the strategy that considers only power-disturbance uncertainty, the proposed method improves the mean value of the control performance standard by 14.98 percentage points and 11.38 percentage points, respectively, and reduces the mean absolute area control error by 5.30 MW and 2.22 MW, respectively, which confirms the effectiveness and superiority of the proposed method. The proposed model and algorithm offers a valuable reference for regional grid frequency control in new-type power systems with high renewable penetration.

Key words: frequency control, model predictive control, inertia time constant, stochastic optimization, uncertainty