Requirements of Power Grid Situation Cognition and Resilience Assessment in Extreme Scenarios

The increase in the number of regulators brings about the problem of high dimensional random variables to the cognition of power network operation state. The input parameter structure and data processing flow of the traditional model are clear and demanding, and the final analysis results are strongly dependent on the institutional model, which does not have the characteristics of model-free and adaptive, and it is difficult to deal with unstructured data or defective data. Through high-dimensional statistical analysis and artificial intelligence technology, the power grid operation state is finely simulated and predicted to improve the power system's adaptability to distributed resource fluctuations, promote the large-scale consumption of distributed new energy, and ensure the stable, reliable and economic operation of the power grid. Through the high dimensional statistical analysis, the high dimensional spatiotemporal big data with low information density is compressed into the low dimensional situation feature with high information density. For extreme scenarios and random fluctuations of massive flexible resources, specific signal tracking/analysis is carried out by adjusting test functions and control hyperparameters to realize specific event cognition, and quantitative assessment of power grid risk and its resilience..