An Optimal SVD Filtering Method for Measurement Accuracy Improvement against Harmonic Disturbance in Grid-Connected Inverters

The increasing penetration of power electronics, such as grid-connected inverters and active loads may cause power quality issues, which reduce the sensitivity of monitoring and control systems due to measurement noises. This article presents an optimal singular value decomposition (SVD) filtering method for grid-connected inverters to improve sampling accuracy against measurement noises. First, the principle of this proposed method is based on the Hankel matrix theory, and then the implementation process is explained, during which the relationship between the Hankel matrix dimension and noise reduction is discussed. Furthermore, the optimal singular value is analyzed and proposed to determine the reconstruction order. Then, the comparative analysis of the proposed optimal SVD filtering method and difference spectrum method is given to explain the optimal reconstruction order. Finally, simulation verifications are implemented to validate the effectiveness of the proposed filtering method, considering the Hankel matrix dimension, reconstruction order, and different signal–noise ratio (SNR). The verification results show that the proposed optimal SVD filtering method can accurately identify the sampling current of grid-connected inverters, even if severe harmonic noises and oscillation happen. The proposed method can reduce the effects of harmonic disturbance on measurement accuracy and control performance of grid-connected inverters, which can improve the robustness of grid-connected inverters.