This paper presents a measurement-based solution for low frequency oscillation(LFO) analysis in both real time monitoring and off-line case study. An online LFO property discrimination method is developed first,which alternately uses empirical mode decomposition(EMD)/Hilbert transform(HT) and square calculation to process the measurement data. The method magnifies the variation trend of oscillating variables to accurately discriminate the property of the oscillation. Subsequently, an oscillation source locating method for the forced oscillation(FO) and a strongly correlated generator identification method for the weak damping oscillation(WDO) are proposed. Finally, numerical study results on a test system of the isolated Changdu grid in Tibet validate the proposed methods.
Estimating low-frequency oscillation modes and the corresponding mode shapes based on ambient data from WAMS measurements has a promising prospect in power system analysis and control.Based on the stochastic subspace method,this paper proposes a revised stochastic subspace method by introducing reference channels,which can estimate the modes and the mode shapes simultaneously with great computational efficiency.Meanwhile,the accuracy of the estimated results is not degraded.To discriminate the real modes from the spurious ones,the stabilization diagram is introduced.A novel algorithm is designed to deal with the stabilization diagram which can detect the real modes automatically.Tests conducted on the IEEE-118 system indicate that the proposed method has good performance in terms of both computational efficiency and accuracy,and has the potential of being used on-line.
