基于Copula传递熵的设备级和网络级宽频振荡传播路径分析及振荡源定位方法

Device and Network Level Wideband Oscillations Propagation Path Analysis and Source Localization Method Based on Copula Transfer Entropy

电力电子化电力系统的宽频振荡问题严重危害电网的安全稳定运行,及时确定宽频振荡的传播路径和振荡源位置,对振荡的抑制至关重要。该文从因果关系的角度提出一种基于Copula传递熵的设备级和网络级宽频振荡传播路径分析及振荡源定位方法,通过分析多变量之间的因果传递方向和因果强度系数,利用有向加权图构建宽频振荡因果网络,在此基础上确定振荡源位置,并通过保留因果强度系数最大的支路,确定振荡的主要传播路径。采用所提方法一方面能够从设备级分析控制器内部各状态变量的因果关联性;另一方面能够从网络级分析电网中各节点振荡数据的因果传递性,从而确定宽频振荡在控制器内部和电网中的传播路径及振荡源位置。最后,在强迫振荡、风机与弱电网交互、发电机轴系振荡中的仿真算例表明,所提方法适用于多种机理的宽频振荡,能够同时实现设备级和网络级宽频振荡传播路径分析和振荡源定位。

Wideband oscillations in the power electronic power system seriously endangers the safe and stable operation of the power grid.Timely determination of the propagation path and source location of wideband oscillations is crucial for suppressing oscillations.Model based analysis methods have limitations such as unknown parameters and high complexity,restaining their wide application in practical power systems.With the development of information theory,the method based on mathematical statistics becomes a promising method for the analysis of wideband oscillations.These methods focus on the correlation between data items and discover the inherent laws of wideband oscillations.Considering that the data of wideband oscillations in the new-generation power system is a multivariable strongly nonlinear and causal time series,it is worth conducting in-depth research on how to analyze the propagation characteristics of wideband oscillations and achieve oscillations source localization from the perspective of causal relationships based on mathematical statistics.Therefore,this article proposes a device and network level wideband oscillations propagation path analysis and oscillations source localization method based on Copula transfer entropy from the perspective of causality.Firstly,the zero-mean normalization method is used to preprocess the analyzed oscillations data.Secondly,the original signal can be decomposed into a series of intrinsic mode components and a residual component using empirical mode decomposition(EMD).The residual component can be removed as it is independent of oscillations.Thirdly,for each mode,calculate the Copula transfer entropy between each variable to determine the direction of causal transfer.Then define the numerical value of Copula transfer entropy as the causal strength coefficient.Fourthly,a wideband oscillations causal network is constructed using directed weighted graphs,where each node in the network is a state variable or bus power signal,and the edge weights between each node are the causal strength coefficients between the corresponding variables.Fifthly,calculate the out-degree of each node in the network,and the node with the highest out-degree is determined as the position of the oscillations source.Finally,starting from the state variable or bus where the oscillations source is located,retain the branch with the highest causal strength coefficient,which is the main propagation path of the oscillation mode.At device level,the causal relationship between state variables within the controller can be revealed,thereby determining the key state variable and oscillations propagation path.At network level,the causal relationship between the power of each bus in the power grid can be revealed,and then the bus where the oscillations source is located and the oscillations propagation path can be determined.On one hand,a simulation example of PMSG connected to grid is used to analyze the causal relationship of oscillations in various links within the controller.The propagation path diagram of the internal oscillations of the PMSG is obtained using the method proposed in this article.The sub synchronous oscillation mode of the system is mainly related to the DC capacitor link,outer voltage loop,and inner current loop of the PMSG.These links form the internal propagation path of this oscillation mode.On the other hand,a simulation example of a four-machine two-area system containing a wind farm is used to analyze the causal relationship of oscillations between the busbars of a complex system.Simulation examples in forced oscillations,interaction between wind turbines and weak grid,and generator shaft system oscillations demonstrate that the proposed method is suitable for various mechanisms of wideband oscillations,and can achieve wideband oscillations propagation path analysis and oscillations source localization from the network level.The following conclusions can be drawn from the simulation analysis:(1)The method based on Copula transfer entropy avoids establishing a detailed model,and its simulation results are consistent with the oscillations propagation path obtained from theoretical derivation.(2)The method is suitable for various mechanisms of wideband oscillations such as forced oscillations,interaction between wind turbines and weak grid,and generator shaft system oscillations.(3)The method can simultaneously achieve wideband oscillations propagation path analysis and oscillations source localization at device and network level.