.png){kind=icon}
.png){kind=icon}
Return
Denoising-Enhanced Compressed Sensing for Power Quality Monitoring: A Unified Framework Integrating Sparse Sampling, High-Fidelity Reconstruction, and Disturbance Extraction
.png){kind=icon}
PRE
AI
Z
Y
H
Z
J
H
Z
DOI：10.1109/TIM.2025.3635808.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
To address the detrimental effects of noise folding (NF) phenomena on compressed sensing (CS) reconstruction performance in power quality (PQ) waveform analysis, this study proposes an integrated algorithm framework encompassing denoising-enhanced compressed sampling, reconstruction, and disturbance component separation through an innovative denoising measurement matrix design and improved algorithm iteration conditions. First, a novel denoising measurement matrix construction method is proposed, which employs 1-D Gaussian convolution kernels for row optimization of the measurement matrix. This critical innovation enables the transformation of conventional noisy CS models into noise-free equivalents through signal-adaptive filtering during compressive sampling, effectively filtering signal noise while preserving full-frequency signaling components. Second, to mitigate the increased correlation between the Fourier orthogonal basis and the identity matrix caused by convolution operations in joint-domain sparse dictionaries, this study constructs an equivalent convolution kernel matrix and proposes a cascaded dictionary structure that combines the Fourier basis with the pseudoinverse matrix. This advancement significantly reduces interatom correlation while enhancing PQ signal sparsity compared with conventional approaches. Third, an enhanced sparse adaptive matching pursuit (SAMP) algorithm is designed, in which a dynamic stopping criterion is introduced on the basis of the weight ratio between terminal and initial atoms in the candidate set. This innovation achieves dual functionality: measurement noise filtration and computational efficiency optimization during reconstruction. Comprehensive validation through simulation tests and field experiments demonstrates that the proposed algorithm exhibits superior performance under low signal-to-noise ratio (SNR) conditions. Notably, in the field of PQ monitoring, this study achieves the first simultaneous implementation of denoising and compressed sampling in noisy environments, significantly improving the compression ratio (CR) while maintaining reconstruction accuracy and enabling the precise extraction of both transient and steady-state disturbance features from PQ signals. These advancements establish a new paradigm for efficient PQ data acquisition and analysis in smart grid applications.
Keywords:
Compressed sensing (CS)
denoising-enhanced compressive sampling
noise folding (NF)
power quality (PQ) disturbance
sparse decomposition
Journal
IF:
5.9
Papers:
1.9W
Citations:
5.8W