Denoising method in fiber optic current transformer based on data characteristics and depth entropy

Abstract

Focusing on the difficulty in suppressing the random error of non-Gaussian noise in fiber optic current transformer (FOCT) measurement, a FOCT noise suppression method based on data modal characteristics and depth entropy is proposed. The empirical mode decomposition model is established, and the signal timing is processed with adaptability, completeness and approximate orthogonality. The measurement signals with high-amplitude interference distributed in each IMF are extracted to solve the problem of modal aliasing interference. The analysis model of the random dynamic characteristics of the time series is established, and the calculation method of the depth entropy is proposed. By calculating the correlation of the time series, the complexity of the information is evaluated, and the signal after filtering the interference is reconstructed. Experiment results show that the average value of the filtered current data is 0.2% deviation from the reference current, and the variance is reduced by 56%. This filtering method can significantly reduce the dispersion of test data while ensuring unbiased estimation. The​ FOCT noise suppression method based on data modal characteristics and depth entropy can filter non-Gaussian noise and the signal-to-noise ratio of FOCT output data can be improved.