Research On Parameter Estimation Method Of Low Frequency Oscillation Signal Based On Interpolation DFT

With the increase of power load and the improvement of power market,the power system has changed from regional self-balance state to large-scale interconnection state.The interconnection of power system can make rational use of energy and bring economic benefits.However,it can increase the complexity of power system dynamic characteristics and cause low-frequency oscillation,which is a threat to system stability and power equipment.The online monitoring and parameter identification of low-frequency oscillation in power system can provide safety warning for power grid,which is of great significance for the safe and stable operation of large power grids.Discrete Fourier Transform(DFT)has the advantages of small calculation amount,high precision and good robustness.It is often used in power system signal parameter estimation,but it is not suitable for oscillation parameter analysis.On the one hand,the oscillation signal is different from the sinusoidal signal model,an d its damping factor cannot be directly estimated.On the other hand,the oscillation signal contains several components with similar frequencies,and the inherent fence effect of DFT makes it impossible to identify them directly and accurately.In this paper,an improved interpolation DFT algorithm suitable for low-frequency oscillating signals of power systems is studied to achieve high-precision estimation of the characteristic parameters of oscillating signals.Firstly,this thesis analyzes the mechanism of low-frequency oscillations and the current status of parameter estimation research.The low-frequency oscillation identification methods can be devided into two categories: analysis methods based on power system models and analysis methods based on system measurement signals.The low-frequency oscillation signal parameter estimation method base on DFT is introduced in detail,and the existing interpolation DFT algorithms are summarized and their advantages and disadvantages are analyzed to provide a basis for the subsequent research.Secondly,according to the current research status of existing interpolation DFT methods,a complex frequency domain trispectral interpolation algorithm based on the Maximum Sidelobe decay Window is proposed in this paper.Compared with the existing interpolation DFT algorithm,this algorithm uses the complex spectrum for interpolation processing,takes into account the interference of negative spectrum,and can improve the accuracy of parame ter estimation.At the same time,the general formula of the algorithm is applicable to all paraflop attenuation windows,which provides a variety of options for practical applications.Then,the thesis analyzes the effect of Gaussian white noise on the accuracy of parameter estimation.Based on the Gaussian white noise,a spectral correlation model of the white noise signal is established and the theoretical variance expressions for the oscillation frequency and damping factor of the three-spectrum interpolation algorithm are derived.According to the theoretical variance expression,it is known that the theoretical variance of parameter estimation is inversely proportional to the Gaussian white noise signal-to-noise ratio and proportional to the square of the sampling frequency.The correctness of the theoretical variance expression is verified by simulation tests in the range of signal-to-noise ratio from 0 to 100 d B.Finally,some simulation experiments about power system oscillation parameter estimation are done in this thesis.The simulation signals are divided into three categories: self-synthesized ideal signal,four-engine two-zone system oscillation signal and grid measured signal.Based on self-synthesizing ideal signals,the proposed interpolation DFT algorithm is compared with three typical interpolation DFT algorithms.The experiment result shows that the proposed algorithm has the highest parameter estimation accuracy in the range of low frequency oscillation frequency(0.1-2.5Hz).In the case of small sampling time,accurate parameter estimation can also be achieved.By analyzing and processing the simulation data of the four-machine two-zone system and the actual measured signal data of the power grid,t he accuracy and effectiveness of the algorithm proposed in this paper are further verified.