Practical Implementation and Operational Experience of Dynamic Mode Decomposition in Wide-area Monitoring Systems of Italian Power System

Abstract

This study presents the assumptions and strategies for the practical implementation of the dynamic mode decomposition approach in the wide-area monitoring system of the Italian transmission system operator, Terna. The procedure setup aims to detect poorly damped interarea oscillations of power systems. Dynamic mode decomposition is a data-driven technique that has gained increasing attention in different fields; the proposed implementation can both characterize the oscillatory modes and identify the most influenced areas. This study presents the results of its practical implementation and operational experience in power system monitoring. It focuses on the main characteristics and solutions identified to reliably monitor the interarea electromechanical modes of the interconnected European power system. Moreover, conditions to issue an appropriate alarm in case of critical operating conditions are described. The effectiveness of the proposed approach is validated by its application in three case studies: a critical oscillatory event and a short-circuit event that occurred in the Italian power system in the previous years, and a 15-min time interval of normal grid operation recorded in March 2021.

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).…”

“…temporal and spatial structures of the modes) using ringdown data. The work [14] analyzes the real implementation of DMD on the Italian power system. Specifically, the authors use the block-enhanced DMD with a sliding window of length 20 s and apply a band filter to estimate only the interarea modes (0.10-0.50 Hz).…”

“…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?…”