Probabilistic assessment of critical inertia in modern power grids via Copula-GAN-based scenario synthesis

The system frequency instability risk in high-penetration renewable energy integrated power systems, caused by inertia deficiency coupled with the stochasticity of grid-connected renewable generation, cannot be accurately quantified by conventional deterministic assessment methods due to the neglect of interdependencies among multi-source uncertainties and impacts of extreme events. This study proposes a two-stage method centered on joint probabilistic scenario generation and dynamic assessment of critical inertia. A Copula-GAN-based joint generation model is first established, where the Gaussian Copula function is employed to constrain the nonlinear correlations between wind/solar power output and load under various fault scenarios. The Generative Adversarial Network (GAN) learns the joint distribution characteristics of extreme events to generate time-series scenarios of wind/solar power output and load. Subsequently, system parameters are adaptively adjusted to tune the equivalent inertia of the system toward the critical inertia threshold. Finally, the probabilistic safety domain of critical inertia is estimated using Monte Carlo simulations. A simulation is conducted on the modified IEEE 39-bus system, and the results demonstrate the effectiveness of the proposed two-stage method.