Research On Online Identification For Low Frequency Oscillation Modes In Power System Based On ARMA Recursive Method

As the development of power grid and the advancement of area inter-connection, adopting a large number of high gain excitation regulators to improve generator voltage accuracy and stability of the system, it makes grid low frequency oscillation(LFO) phenomena appear often so that it has serious threat for the system runs normally.Therefore, LFO has become a restricted factor for theregional network inter-connection, accurate and timely low-frequency oscillation mode identification has the vital significance for the safe and stable operation of the power grid. As the wide-area measurement system widely used in power system, it provides a real-time data platform for the online dentification of LFO mode. Therefore, one kind method which is not dependent on system model but only based on the measurement signal of the system become the research hotspot in domestic and foreign.Firstly, this paper summarizes three elements for the identification of LFO modes based on the measured signal from the angle of system identification: ambient signal, autoregressive moving average (ARMA) model and the ARMA recursive algorithm for solving the model parameters; and also discusses the feasibility of using ambient signal to identify the LFO modes in detail, the related basic theory of ARMA model amd the calculation principle for LFO modes parameters.Secondly, based on conditional ARMA recursion algorithm of the ARMA model parameters estimation, the improved algorithm uses the obtained white noise estimates by fitting the higher order autoregressive (AR) model by the weighted recursive least square (WRLS) method in the conventional WRLS method, and then it has the preferable accuracy and the fast convergence rate of parameter identification. Combined with the three key elements of LFO modes identification, this paper also puts forward a total method of LFO modes identification which contains five parts: selection of the ambient signal, data pretreatment, determination of the order of the AR and ARMA model, the estimation of the ARMA model parameters, the calculation of the LFO modes and the extraction of the major modes.Finally, this paper makes the time domain simulation simulate the ambient signal of the power system at the disturbance of the white-noise by using the MATLAB and Power System Analysis Toolbox, and makes the LFO modes identification at the condition of stable state and the condition of the large disturbance occurring. The results prove the availability and the effect of improved algorithm. What is more, an active power signal of a transmission line of the southern power grid is also adopted to estimate the LFO modes, and comparison with the results of the ringdown signal by Prony algorithm analysis shows the effect of the identification method. Analysis of the algorithm principle and the calculation speed of the examples show that the proposed LFO method is suitable for online application.