Synchronized Data-Based Identification of Electromechanical Oscillation Modes Using Improved FastICA Considering Measurement Noise

Synchrophasor-based identification of low-frequency oscillation is an effective means for monitoring the security and stability of power systems, while the loss of identification accuracy due to measurement noise remains a major obstacle. This article proposes an improved fast independent component analysis (improved FastICA) algorithm to identify oscillation modes (oscillation frequency and damping ratio) from noise-contaminated synchronized measurements. To improve the ability of conventional FastICA to extract modes from noise-polluted synchrophasor, nonlinear matching pursuit (NMP) is introduced to replace the negative entropy calculation of mono-frequency modes in FastICA. NMP enables FastICA to identify oscillation frequency and damping ratio even under strong noise conditions, thereby significantly improving identification accuracy. The performance of the proposed method is evaluated on synthetic signals and numerical simulation signals from the IEEE 16-machine system. The results of the identified modes in all cases, as well as comparison results from conventional FastICA, empirical mode decomposition (EMD), and variational mode decomposition (VMD), all combined with Hilbert transform (HT), confirm the accuracy, robustness, and efficiency of the proposed method for oscillation mode identification.