An interactive tool for measurement-driven modal analysis of large-scale power systems

Abstract

The desire to reduce model-dependency has driven measurement-based modal identification methods, as an alternative to analyzing linearized system models. One such measurement-based method is Dynamic Mode Decomposition (DMD), which has recently appeared in literature. Using this relatively fast DMD algorithm, this paper presents an interactive modal-identification tool for spatio-temporal analysis of measurement data. The tool can automatically scan through measurements, and display the values of oscillation frequency and damping ratio, as well as reconstruct signals. The use of this tool, its options, and visualization capabilities are illustrated using simulated measurements from an interconnected power grid.

An improved dynamic mode decomposition for real‐time electromechanical oscillation monitoring in power systems: The impact of ultra‐low frequency modes and its removal strategy

“…DMD is one of the data-driven methods that has the potential to be used in the power system application due to its robustness and estimation speed. Specifically, DMD has been used for electromechanical oscillation monitoring [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] estimation of inertia [17] and frequency [18,19], harmonic distortion monitoring [20,21], fault detection [22], load forecasting [23][24][25] and data-driven control for enhancing rotor angle stability [26,27]. Regarding the oscillation monitoring using DMD, there are still some challenges related to the amount of input data that need to be addressed for achieving real-time oscillation monitoring.…”

“…Feature addressed Target Data type New technique proposed [3][4][5] Modal identification ability Analysis Ringdown None [6] Robustness [7,8] Computational speed [9] Accuracy [2], [10][11][12], [14][15][16] Robustness Improve optimization, data stacking, randomization, data processing, energy evaluation [13] Accuracy Non-linear mapping FIGURE 1 The behaviour of speed signal during real-time operation (ambient and ringdown data).…”

“…ULFMs can also be observed in Figure 9 of [11]. However, the characteristics of ULFMs and trends contained in the ringdown data are not discussed and analysed in [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16].…”

“…In addition, it is still unclear if (and how) we can further reduce the window length as it will increase the computational speed. Therefore, in comparison to our previous work in [9] and the literature presented in [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16], this paper proposes a detrending strategy as a preprocessing step where the measurements from the PMUs are passed through a high-pass IIR filter. This trend elimination improves the DMD's accuracy for the considered three time-window lengths and as a result the time-window length can be reduced up to 4 s. In addition, this paper addresses the following open yet fundamental questions regarding the ULFM: What are the components in the power system which excite the ULFMs?…”