Bidirectional frequency-domain digital filtering to simultaneously improve temperature resolution and eliminate spatial inaccuracy of a distributed temperature sensor

Abstract

This article discusses a digital filtering technique to improve the temperature resolution of a Raman fiber optic distributed temperature sensor. A frequency-domain digital filtering algorithm has been used for this typically time-domain task to show that by reducing the effect of noise introduced by analog electronics and the quantization noise of the analog-to-digital converter in the detection and data acquisition stages, superior temperature resolution is achievable. An important feature of the filtering scheme used is that no spatial inaccuracy is introduced into the ranging of hot zones, despite the highly nonlinear phase response of the filter. Digital filtering used for postprocessing of data shows clearly and unambiguously a temperature resolution of 5 K in the present setting of our experiment, with the possibility of much greater improvement. Merely using analog signal conditioning does not provide the same clarity and uniqueness in temperature resolution and spatial location that digital filtering offers. This digital filtering scheme offers greater flexibility and versatility than mere signal-averaging approaches to improve the SNR of such systems. The filter's ability to reject deterministic interfering frequencies of appreciable energy content is also demonstrated by simulating such frequencies of much greater energy than the actual signal.