基于不完全WAMS信息的电力系统暂态稳定评估与紧急控制

The researches on the foundational theory and key technology of the large power system transient stability assessment and emergency control under incomplete WAMS environment will be taken in this project. The post-fault generator rotor angles and terminal voltages are regarded as a whole cluster, the global fault characteristics are picked up through feature definition and choices, the power system transient stability online prediction and assessment are performed based on machine-leaning approaches. The Principal Component Analysis will be conducted on the post-fault generator rotor angle cluster, the virtual generator will be constructed using the principal component and its weight coefficients, the transient stability will be evaluated based on equal-area principle, the emergency controls will be obtained. The main innovation of this application is the integration of the power system transient stability prediction, assessment and emergency control under the incomplete WAMS environment, the effective way to apply equal-area principle under such environment will be found. All the models, algorithms and control strategies proposed in this project have strong anti-disturbance ability and robustness, do not rely on the system topology and model parameters, prompting the power system transient stability theory and technology to practical industrial application, and enhancing the power system transient stability.

本项目对基于不完全WAMS信息的电力系统暂态稳定评估和紧急控制的基础理论和关键技术开展研究。将故障后发电机转子角和机端电压轨迹簇视为一个整体，通过特征定义和筛选提取广域故障特征，采用智能机器学习方法进行电力系统暂态稳定的在线预测和稳定程度的分类评估。对发电机转子角轨迹簇信息进行主成分分析，根据提取的主成分和权重系数构建虚拟发电机，应用等面积法则进行暂态稳定评估，根据加减速面积差额确定相关发电机的切机量，给出优化组合的紧急控制策略。本项目最大的创新是在不完全WAMS信息条件下实现了电力系统暂态稳定在线预测、定量评估和紧急控制的一体化，找到了不完全WAMS信息条件下应用等面积法则的有效路径。所提方法对不完全WAMS信息具有极强的抗干扰能力和鲁棒性，不依赖于系统网络结构和模型参数，从而将基于响应的电力系统暂态稳定评估和控制的理论和方法推向实用化，增强电力系统安全稳定控制的第二道防线。

本项目面向不完全WAMS信息条件下和人工智能建模方法的电力系统暂态稳定在线预测，针对输入特征、人工智能算法在电力系统暂态稳定中的应用以及预防控制展开研究，提出了不随系统规模变化而变化的轨迹簇特征，建立了适合大规模交直流互联网的暂态稳定评估模型，并且研究了多类型自适应组合分类器以及迁移学习在暂态稳定方面的应用。