Empirical mode decomposition based dynamic error correction in SS covered 62.5/125 μm optical fiber based distributed temperature sensor

Abstract

The design and implementation of empirical mode decomposition (EMD) based preprocessor for backscattered spontaneous Raman anti-Stokes (AS) and Stokes (St) signals obtained from a stainless steel (SS) covered, 62.5/125 μm optical fiber based distributed temperature sensor is presented. The preprocessor dynamically minimizes the error in temperature measurement caused by the difference in attenuation to AS and St signals offered by the optical fiber. Simultaneous denoising of AS and St signals obtained by the EMD based preprocessor yields better signal to noise ratio (SNR) of these signals and allows reduced error in temperature measurement. The EMD based technique is much better than previously reported techniques in terms of simplicity and automation. Automated and dynamic self calibration of distributed temperature sensor is also possible with the proposed preprocessor in an easier way. The use of proposed preprocessor has been demonstrated to develop an optical fiber based distributed temperature sensor with an accuracy of ±2.5 °C in a temperature range of 25-105°C over a sensing length of 90 m with a spatial resolution of 1 m. The developed system uses a rugged stainless steel (SS) covered 62.5/125 μm Multimode optical fiber. SS covering on the fiber makes it easier and safer to install the sensing fiber in critical field locations where normal sensing fiber cannot be used. © 2014 Elsevier Ltd.