Phase Noise Suppression in Phase-Sensitive OTDR Using Fault-Tolerant D-S Evidence Theory

The phase noise caused by various factors will lead to the distortion of phase demodulation in phase-sensitive optical time domain reflectometer. In order to suppress the phase noise, a data fusion technology based on fault-tolerant D-S evidence theory is introduced for phase demodulation. The multi-sensor feature layer dataset is constructed after the cubic spline interpolation to obtain three sets of original phase data. The Kalman filter removes phase difference noise for the optimal state estimation, and the chi-square detection uses residuals to exclude abnormal phase data. The D-S evidence theory is then combined to realize the weight allocation and phase reconstruction. Experimental results show that fault-tolerant D-S evidence theory can realize the stable and accurate phase reconstruction of sinusoidal and harmonic vibration signals in the range of 5 Hz∼200 Hz at about 42 km position of optical fiber. The SNR values of reconstructed phases after data fusion are largely improved after signal noise suppression in the amplitude spectral densities diagrams and power spectral densities diagrams. The repeatability test shows a correct rate of reconstructed phase up to 90%, which confirms the better stability and accuracy of phase demodulation. Finally, the feasibility in actual industrial field is verified by the phase reconstruction of vibration motor signal.