电力系统参数空间暂态稳定边界构建及在线快速更新方法

Construction and Online Fast Update Method for Transient Stability Boundary in Parameter Space of Power System

为适应在线暂态稳定分析与控制,提出了一种电力系统参数空间中的暂态稳定边界构建及快速更新方法。对于给定的关键故障和既定故障前后的网络拓扑结构,首先,基于横向可扩展的宽度学习系统,构建了极限切除时间与电力系统参数之间的映射关系。为提高实际故障切除时间阈值附近的预测准确率,通过构建二次比例因子对临界误差进行修正。然后,结合预设故障切除时间阈值确定相应暂态稳定边界,并评估当前系统的稳定裕度。最后,在保证暂态稳定评估准确性的基础上,基于增量学习方法提出了无须重新训练全部网络的在线快速更新策略。通过对IEEE 39节点测试系统和中国南方电网实例系统仿真分析发现,所构建模型能够对系统暂态稳定性进行准确评估,并具有良好的泛化性能。同时,快速更新策略可在保证预测准确率的情况下,大幅减少模型更新时间,为在线暂态稳定评估提供了支撑。

To adapt to online transient stability analysis and control,this paper proposes a construction and fast update method of transient stability boundary(TSB)in the parameter space of the power system.For the given critical fault and the network topology before and after the given fault,the mapping relationship between the critical clearing time and the power system parameters is constructed based on the horizontally scalable broad learning system.In order to improve the prediction accuracy near the actual fault clearing time threshold,the critical error is corrected by constructing a quadratic scaling factor.Combined with the predefined fault clearing time threshold,the corresponding transient stability boundary is determined,and the stability margin of the current system is evaluated.Finally,on the basis of ensuring the accuracy of transient stability assessment,an online fast update strategy without retraining the entire network is proposed by using the incremental learning method.Through the simulation analysis of the IEEE 39-bus test system and an actual system of China southern power grid,it is found that the constructed model can accurately assess the transient stability of the system and has good generalization performance.Meanwhile,the fast update strategy can greatly reduce the model update time while ensuring the accuracy of the prediction,which provides support for the online transient stability assessment.