基于循环神经网络的动态等值模型辨识

为了在系统稳定计算分析中简化和减小计算难度,用动态等值作为一种建模方法提出了一种基于循环神经网络的电力系统动态等值法并结合串并联辨识结构将其用于等值后模型的辨识。这种方法不需要预先建立确定的动态模型,仅靠边界节点的测量数据。循环神经网络的权值和结构决定了被等值外部模型的参数和结构。对IEEE的39节点进行的仿真结果表明了该方法能够捕捉原始系统的动态特性,有较高的精确性。

Recurrent Neural Network-based Dynamic Equivalencing and Identification

In stability analysis,only particular areas need to be further detailed simulation and analyzed.It is espected that systems can be simplified and without employing great programming effort or extensive time-consuming computations.Dynamic Equivalent as a modeling method can reduce the huge workload when be used in analyzing the large interconnected power system.On the other hand,recurrent artificial neural netwoks(ANN) has good ability to deal with the complicated nonlinear problem and series parallel model(SPM) can assure the stability and astringency of the identified model.In this paper,dynamic equivalent for transmission networks in the interconnected power systems using the recurrent artificial neural netwoks(ANN) is introduced and series parallel model(SPM) structure is chosen to identify equivalent model.The approach depends on measurements of the boundary nodes,need not to postulate any fixed dynamic model in advance.The weighting parameters and structure of ANN will define the parameters and structure of the external areas.This new approach is used to simplify the 39 nodes network of IEEE,the results show that the proposed method can capture the dynamic character of the original system,and show the accuracy of the equivalent model and the validity of the proposed method.