Short-time and weak signal demodulation for fiber optic current sensors based on backpropagation neural network

It has been challenging to demodulate short-time and weak current signals collected by fiber optic current sensors (FOCSs) under ultra-high voltage, since the background noise can significantly affect the spectra of the current signals. To address this issue, here we propose a novel FOCS demodulation method based on back propagation neural network, where the impact of the noise on the measurement can be largely reduced. The demodulation method can determine the amplitudes of short time series of weak currents with high resolution and accuracy. To evaluate the performance of the proposed method, the demodulation accuracy and robustness of the method are experimentally investigated and compared with those of the traditional fast Fourier transform (FFT) method. The experimental results show that our method can produce reliable results when demodulating weak current signals over short time windows of less than one period, and achieve a significantly lower standard deviation (6.7 mA) compared with the FFT method (15.6 mA) in the current range 0-0.306 A. The higher robustness against the background noise than the FFT method and the excellent repeatability of our demodulation method are also demonstrated through simulations. These results suggest that the proposed method will provide an effective way to improve the detection performances of FOCSs in fast dynamic measurements.

Automated summary

A novel FOCS demodulation method based on back propagation neural network is proposed to address the challenge of demodulating short-time and weak current signals under ultra-high voltage. The proposed method reduces the impact of background noise and achieves high resolution and accuracy in determining the amplitudes of weak currents. Experimental results demonstrate the reliability and robustness of the method compared to the traditional FFT method.