数据-物理联合驱动的大电网频率安全智能评估

Joint Data-Physical Driven Frequency Security Intelligent Assessment of Large Power System

新型电力系统的低惯量特征为其频率安全带来严峻挑战。为快速准确评估大电网受扰后的频率响应,文中提出数据-物理联合驱动的大电网频率安全智能评估方法。为实现数据模型和物理模型分析手段的有效结合,设计了频率安全评估的可信集成学习方法,准确评估数据驱动频率安全结果的可信度,并通过设定可信度阈值作为数据模型和物理模型的切换依据。若数据模型评估结果的可信度高于阈值,则采纳为可靠的数据驱动频率安全评估结果,否则切换为基于物理模型的时域仿真方法进行评估。利用四川电网仿真模型生成数据集并进行模型性能分析,结果表明所提方法兼具频率安全评估的快速性和准确性。

The low-inertia characteristic of new power system brings serious challenge for its frequency security.In order to quickly and accurately assess the frequency response of large power grids after disturbance,a joint data-physical driven intelligent assessment method for frequency security of large power grids is proposed.To realize the effective combination of data model and physical model,a credible ensemble learning method for frequency security assessment is designed to realize the accurate credibility evaluation of data-driven frequency security results,and the credibility threshold is set as the basis for switching between data model and physical model.If the credibility index of data model is higher than the threshold,it is adopted as a reliable data-driven frequency safety evaluation result,otherwise it is switched to the physical model-based time-domain simulation method for evaluation.The simulation model of Sichuan power grid is utilized to generate datasets and carry out model performance analysis,and the results show that the proposed method combines the rapidity and accuracy of frequency safety assessment.