High Accuracy and Real-Time Positioning Using MODWT for Long Range Asymmetric Interferometer Vibration Sensors

Abstract

The key technology to demodulate the vibration position along the sensing fiber link in the asymmetric fiber interferometer based vibration sensors is to extract similar variation features with high efficiency, so that the asymmetry of the vibration sensors can be effectively eliminated. The wavelet-based approaches have been developed and demonstrated as effective methods for the time-frequency feature analysis of non-stable signals. In this article, an ameliorated real-time positioning algorithm using maximal overlap discrete wavelet transformation (MODWT) for the long range asymmetric fiber interferometer based vibration sensors has been proposed. Firstly, a median filter is employed to remove noise and to acquire the endpoint of the vibration signal. Thus a much higher endpoint detection precision can be obtained. Secondly, the time-frequency features are acquired through the MODWT, which can further improve the time-frequency distribution resolution and processing efficiency. Thirdly, the time delay between features is obtained using a cross-correlation algorithm. Finally, the vibration position information is demodulated based on the time delay value. The performance of the proposed wavelet-based algorithm was assessed and compared to previous studies in an asymmetric dual laser Mach-Zehnder interferometer based vibration sensing system. And the proposed scheme presented good results in both the positioning accuracy and the efficiency in field tests. Specifically, a detection of 98.2% of positioning errors are distributed within the range of ±20 m at a sensing length of 82 km and a mean processing time of 135 ms is also achieved. Therefore, the proposed scheme can expand the areas of application fields for long range vibration sensing.