Inverse Uncertainty Quantification Assisted Forward Uncertainty Quantification in Power System Dynamic Simulations

Abstract

Forward uncertainty quantification (UQ) in power system dynamic simulations is gaining increasing attention today because it assesses the random impacts on dynamic behaviors caused by renewables, loads and model parameter errors. To precisely quantify these random impacts, the prerequisite relies on the correct modeling of input uncertainties. This is possible for loads and renewables, as their uncertainties can be directly obtained from the measured data. However, for transient parameters, such as inertia, damping ratio, and control gains, whose values cannot be directly measured, no existing forward UQ method can accurately capture their uncertainties. As no surprise, a Gaussian or uniform distribution is typically assumed for convenience, which inevitably sacrifices the accuracy of the forward UQ. To address this issue, we propose, for the first time, an inverse UQ-based framework to provide accurate transient parameter uncertainties that assist forward UQ in achieving better accuracy. Furthermore, a decentralized strategy is adopted to improve the scalability of the proposed framework. The simulation results demonstrate that the proposed framework can accurately obtain the forward UQ results for transient parameters by systematically capturing all statistical information about their uncertainties, outperforming the commonly used empirical method.