Research On Instantaneous Frequency Identification Algorithm Of Power System Transient Signal

Frequency is a critical physical quantity of power system,which reflects the operation status of the system and is an important basis for monitoring,protection and control.The instantaneous frequency of the transient signals in power system is used in frequency and voltage regulation automatic control device,under-frequency load-shedding device,system automatic disconnection device,and power system relay protection device,etc.In the past,the frequency identification algorithms are focus on the steady-state process of power system,and many of them have achieved satisfactory result.However,during transient process of power system,the frequency of the system shows time-varying characteristics,which makes the carrier frequency signals,such as voltage signal,show non-stationary characteristics.The traditional frequency identification methods based on stationary signal assumption is no longer suitable.The instantaneous frequency to be identified is considered from the perspective of non-stationary signal analysis in this paper.Time-frequency analysis methods are introduced to obtain the time-varying characteristics of non-stationary signals.High concentration of the time-frequency distribution can accurately reflect the dynamic behavior of signals and obtain the instantaneous frequency value.This paper introduces the related concepts of time-frequency analysis and some traditional time-frequency analysis methods,and analyzes their characteristics and application scope.It focuses on the synchronous compression transform based on short-time Fourier transform and polynomial chirplet transform based on parametric time-frequency analysis.From the perspective of inner product theory,we analyzed the principle of two kinds of methods to obtain the high concentration time-frequency distribution of non-stationary signals,and used an example to show the performance of the methods.We proposed an instantaneous frequency identification algorithm,called “Parameterized Frequency Undetermined Coefficient method(PFUC)” for non-stationary signals.The algorithm is based on the relationship between the signal phase and time,and uses Taylor's expansion to parameterize the time-varying frequency.The simulation results show that when the frequency is modulated by linear or nonlinear polynomial function,the algorithm can achieve the goal of higher instantaneous frequency identification accuracy,which is better than Wigner Ville transform and Hilbert transform.In addition,the algorithm has the advantages of high efficiency and no influence of the signal sampling rate.Improving the signal sampling rate can improve the calculation accuracy but not change the calculation time.Considering the time-varying frequency of power shortage,the instantaneous frequency and its carrier frequency are modeled respectively.Accurate identification of the instantaneous frequency of this kind of signal can provide the operation basis for the under-frequency load-shedding device.Its frequency time-varying feature is that it changes with the negative exponential law of time,and may be accompanied by a certain degree of oscillation.The second-order and fourth-order synchrosqueezing transform based on short-time Fourier transform are used to analyze the negative exponential frequency change signal.The time-frequency distribution with high concentration can be obtained and the frequency time-varying law of the signal can be accurately reflected,but the accuracy of instantaneous frequency value identification is not accuracy enough.The negative exponential frequency can be accurately identified by polynomial chirplet transform,but the calculation accuracy is affected by many parameters such as window function width,and the calculation time is high.The proposed algorithm,PFUC,can accurately identify the negative exponential change frequency and that with oscillation change,and the performances of calculation time and accuracy performance are better than that of synchrosqueezing transform and polynomial chirplet transform.In addition,the PFUC algorithm is not sensitive about amplitude time-varying behavior of the carrier frequency signal.The change rate of frequency can also be used as the action basis of automatic under-frequency load-shedding device.Polynomial chirplet transform and the PFUC algorithm define the higher derivative of frequency,which can be used to calculate the rate of change of frequency,that is,the first derivative of frequency.The simulation results show that the PFUC algorithm can accurately identify the change rate of negative exponential frequency and that of the negative exponential frequency with oscillation,and the accuracy is better than the results of the polynomial chirplet.However,the magnitude of the error is higher than that of the instantaneous frequency identification results.Considering the time-varying frequency under the fault disturbance,the frequency fluctuates near the rated value.The instantaneous frequency is represented by a function with sine factor.Accurate identification of this kind of instantaneous frequency can provide action basis for automatic control device,automatic disconnection device and relay protection device of power system.Simulation results show that the PFUC algorithm can accurately identify this kind of instantaneous frequency,and the accuracy is better than Wigner Ville transform and Hilbert transform.What's more,the voltage signal of short-circuit process usually contains DC component with aperiodic attenuation.In this paper,Empirical Mode Decomposition(EMD)is used to preprocess the signal,and the effect is good.Finally,this paper summarizes the problems of time-frequency analysis methods and the PFUC algorithm that need to be improved.