High-Resolution Raman-OTDR Distributed Temperature Sensors Based on Fast-Non Local Means Denoising Algorithm

Raman-optical time domain reflectometry (R-OTDR) -based distributed temperature sensing (DTS) with high temperature resolution is of great significance in extremely harsh environment of deep wellbore. In this work, we experimentally demonstrate a Fast-non local means (Fast-NLM) algorithm for performance improvement without modifying the hardware architecture of a conventional R-OTDR system. In the proposed method, the first-in-first-out (FIFO) data transmission protocol is used to reconstruct the two-dimensional (2D) temperature image in real time. An integral image is employed to simplify the calculation of Euclidean distance between similarities in the 2D space-time domain. The overall response time by using the method decreases from 27.9 s to 1.7 s, marking a 16-fold enhancement over traditional NLM denoising methodologies. An average temperature resolution is below 0.01 °C at the end of fiber. The evolution of the measured temperature at the hot spot position does not produce delay or distortion. Moreover, two segments fiber at 300 m and 9.68 km are rigorously tested with the temperature range from 50 °C to 300 °C and time span of 4 hours, compared with conventional wavelet denoising algorithm, the root means square error decrease from 1.02 °C to 0.42 °C. In all, the Fast-NLM denoising based R-OTDR illustrates great improvement in temperature resolution while maintaining fast response time. This achievement holds immense potential for diverse applications, including dynamic monitoring in the realms of energy development, oil and gas exploration, and allied fields.