The Studying and Simulating Analysis of the Power Harmonic Measuring Method Based on the Blind Source Analysis

Songlin Guo, Jingyu Wang, F. LiuInternational Journal of Signal Processing Image Processing and Pattern Recognition 2013 · DOI

Abstract

According to electric power harmonic measuring method of the digital sampling, the simulating was carried out in this paper, thus a kind of based on Newton iterative blind source analysis of Fast -three times the ICA algorithm was Put forward, and its convergence was proved, the simulation analysis of the real-time error and measurement algorithm was finished, its instantaneity and measuring of the algorithm was proved.

Introduction

Modern power systems are increasingly populated with nonlinear electronic devices, such as variable frequency drives and switching power supplies. While efficient, these devices inject "harmonics"—distortions of the standard 50Hz/60Hz sine wave—into the grid. These harmonics degrade power quality, cause equipment overheating, and lead to significant measurement errors. Traditional detection methods like the Fast Fourier Transform (FFT) often suffer from spectral leakage, while more precise methods like Mallat Wavelet analysis are computationally too heavy for real-time monitoring. This paper proposes an optimized approach using Blind Source Separation (BSS) to achieve high accuracy with real-time performance.

Methodology: Improved Fast-ICA

The researchers focused on Independent Component Analysis (ICA), a statistical technique that recovers hidden signals (harmonics) from a mixture without prior knowledge of the mixing process.

The core innovation is the replacement of the standard second-order Newton iteration with a third-order Newton iteration method. The process involves:

  1. Preprocessing: Centering and whitening the input signal to remove correlations and normalize variance.
  2. Negative Entropy Criterion: Using negative entropy as a measure of non-Gaussianity to identify independent source signals.
  3. Third-Order Convergence: By modifying the iterative update rule, the authors mathematically proved that the algorithm reaches a solution in significantly fewer steps than the standard version.

Simulation and Signal Analysis

To validate the algorithm, the researchers utilized synthesized data representing various grid conditions. This included stationary mixtures of 3rd, 5th, and 7th harmonics, as well as non-stationary signals designed to test the algorithm's robustness.

Synthesized non-stationary input signal with a frequency jump used for algorithm validation.

As shown in the figure above, the algorithm was tested against signals with abrupt frequency shifts (e.g., jumping from 50Hz to 350Hz) to ensure it could handle real-world discontinuities without losing tracking accuracy.

Key Results and Discussion

The primary advantage of the third-order Newton iteration is its convergence speed. In comparative tests, the improved algorithm achieved an average 37% reduction in processing time over the standard second-order approach. For instance, in one experiment, processing time dropped from 0.8386s to 0.5263s.

Separation of fundamental and harmonic components using the improved Fast-ICA algorithm.

Beyond speed, the separation quality was exceptionally high. The figure above illustrates the successful isolation of the fundamental wave and its associated harmonics, even when subjected to complex phase shifts and amplitude variations. Visual error analysis further confirmed that the improved algorithm maintains a much lower and more stable error curve compared to the original Fast-ICA.

Conclusion

The proposed third-order Newton iterative Fast-ICA algorithm provides a practical "Goldilocks" solution for power grid monitoring: it offers the precision of advanced wavelet methods while maintaining the speed necessary for real-time industrial deployment. Future work will focus on testing this algorithm on low-power embedded hardware and validating its performance against high-frequency stochastic noise in actual industrial power grid environments.

Summary updated

Abstract

According to electric power harmonic measuring method of the digital sampling, the simulating was carried out in this paper, thus a kind of based on Newton iterative blind source analysis of Fast -three times the ICA algorithm was Put forward, and its convergence was proved, the simulation analysis of the real-time error and measurement algorithm was finished, its instantaneity and measuring of the algorithm was proved.
BibTeX
@article{guo2013studying,
  author = {Songlin Guo and Jingyu Wang and F. Liu},
  title = {{The Studying and Simulating Analysis of the Power Harmonic Measuring Method Based on the Blind Source Analysis}},
  journal = {International Journal of Signal Processing Image Processing and Pattern Recognition},
  year = {2013},
  doi = {10.14257/ijsip.2013.6.6.08},
}
Figure