SNR enhancement for Raman distributed temperature sensors using intrinsic modal functions with improved adaptive wavelet threshold denoising

Abstract

To solve the low signal-to-noise ratio (SNR) and loss of signal detail in conventional denoising algorithms for Raman distributed temperature sensors, a novel approach is proposed. This method utilizes improved adaptive wavelet threshold denoising and Pearson correlation coefficient for processing intrinsic modal functions (IMFs). IMFs are obtained by complete ensemble empirical mode decomposition of the original signal, and the correlation coefficient of each IMF is calculated to identify its constituent components. The purely noisy IMFs are discarded, while those containing both noise and temperature signals undergo improved adaptive wavelet threshold denoising, and IMFs containing solely temperature signals are fully preserved. The final signal reconstruction is performed using the IMFs that have undergone noise reduction and the IMFs containing only temperature signals. Experiments show that SNR improvement from 12.83 dB to 19.04 dB at the 25 km fiber, with a reduction in maximum temperature error from 0.95 °C to 0.47 °C, and requires less cumulative averaging time.