AI 中文总结
该研究针对现有模型难以重构电网频率分布复杂非高斯特征的问题，提出结合超统计量与最大似然估计的方法，仅用频率数据即可高保真推断不同特性电网的频率分布，为电网稳定控制提供替代方案。
AI 中文摘要
电网频率的精密测量可追踪能源供需失衡，持续波动表明可能导致停电的压力，因此表征电网频率波动的统计特性对实现高效控制与稳定运行至关重要。现有模型在重构观测到的频率分布的复杂非高斯特征方面成效有限。本文提出一种用于频率波动的可解释随机过程统计推断方法，通过粗粒度的功率失衡信号结合非线性发电机控制与高斯白噪声对频率建模。我们开发了一种结合超统计量的最大似然估计高效算法来推断这些潜在变量。我们在来自英国和南非的大型新测量数据集上测试了该方法，尽管这些电网具有显著不同的特性，但我们对频率分布的预测与测量结果匹配度极高。我们的方法仅使用频率数据进行推断，因此为数据密集型方法提供了一种替代方案。
英文摘要
Precision measurements of power grid frequency track energy supply-demand imbalances, with persistent fluctuations indicating strain that can lead to outages. Characterizing the statistical properties of grid frequency fluctuations is therefore essential to achieve efficient control and stable operations. Existing models had limited success in reconstructing the complex non-Gaussian features of observed frequency distributions. Here, we introduce a method for statistical inference of an interpretable stochastic process governing frequency fluctuations, modeling frequency through a coarse-grained power imbalance signal combined with nonlinear generator control and Gaussian white noise. We develop an efficient algorithm to infer these latent variables via maximum likelihood estimation, combined with superstatistics. We test our method on a large dataset of new measurements from Great Britain and South Africa. Although these grids have markedly different properties, our predictions for frequency distributions match measurements excellently. Our method uses only frequency data for inference, thus offering an alternative to data-intensive approaches.
Comments13 pages, 9 figures, 1 table
