Online Monitoring Of Low Frequency Oscillation And Location Of Distuebance Source Based On The IAMD Method

Because of the large range interconnection and the large scale access with new energy sources of power grid,the condition and structure of power system have changed a lot.Power system is transforming to a new energy power system.The main feature of the new energy power system is "double high",that is,the ratio of power electronic equipment and new energy sources are higher.The dynamic characteristics of “double high” power system have changed profoundly.At the same time,it brings lots of new stability problems.Among them,the probability of low frequency oscillation is increasing and the form is more complex.There are new types of low-frequency oscillation,such as intensive low-frequency oscillation and intermittent low-frequency oscillation.All of these new oscillations will endanger the safety of power grid.Therefore,it is still important to identify new oscillations parameters and locate the disturbance source.When studying low-frequency oscillation,the first problem is to identify the oscillation parameters accurately.In order to improve the accuracy and rapidity of parameter identification,this paper propose to use the analytical mode decomposition(AMD)method.Firstly,the signal is analyzed by FFT.Secondly,the dominant mode determination criteria are given to determine the dominant mode frequency according to the characteristics of on-line monitoring of low-frequency oscillation.Thirdly,the mean value of the adjacent dominant mode frequency is taken as the truncation frequency,and then use AMD method to decompose the signal to obtain the single frequency signal.Finally,the frequency and attenuation factor are obtained by Hilbert transform and least square method.The simulation results confirm that the AMD method is suitable for on-line analysis and monitoring of typical low frequency oscillation signals.This paper has further improved the function of AMD method for parameter identification of new low-frequency oscillation(intensive low-frequency oscillation and intermittent low-frequency oscillation),and the core problem of improvement is to determine the optimal cut-off frequency of AMD decomposition.Firstly,this paper construct auxiliary signals according to the estimated frequency and amplitude through FFT,and then calculate the Euclidean distance between the decomposed single frequency signal and the auxiliary signal as the basis for selecting the optimal truncation frequency;secondly,use chaotic particle swarm optimization(CPSO)to find the frequency corresponding to the minimum Euclidean distance,which is the optimal truncation frequency.Finally,according to the optimal truncation frequency,the original signal is decomposed by AMD to get a single frequency signal,and the parameters are identified.The parameter identification accuracy of the improved analytical mode decomposition(IAMD)method is significantly improved which can realize the accurate identification of the parameters of intensive low-frequency oscillation and intermittent low-frequency oscillation.The location of forced oscillation disturbance source is another important problem in the research of low frequency oscillation.This paper propose a new method to locate the disturbance source more accurately.Firstly,use the IAMD method to obtain the active power fluctuation and voltage frequency fluctuation components which are strongly related to the oscillation.Secondly,calculate the energy function.Finally,support vector machine(SVM)is used to locate the disturbance source accurately.Simulation and examples confirm that the combined method can quickly and accurately locate the disturbance source of forced oscillation.