A near‐infrared trace hydrogen sulfide sensor based on CEEMDAN and PSO‐LSSVM

Abstract

A distribution feedback laser sensor for high precision and high sensitivity detection of the hydrogen sulfide in associated gas from oil fields is developed in this paper. Tunable diode laser absorption spectroscopy and wavelength modulation spectroscopy and were utilized for the H 2 S concentration detection in the oilfield. Complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) algorithm is used to remove the noise from the spectral signal. Particle swarm optimization-least squares support vector machine (PSO-LSSVM) model is applied for concentration prediction. The root mean square error of long-term stability is 0.323. Allan-Werle deviation analysis shows that the theoretical limit of detection is 71 parts per billion while the optimal integration time reaches 96 s. The sensor is of great significance for realtime detection of hydrogen sulfide concentration in associated gas in oil fields.

Real-time methods of hydrogen sulfide detection

“…However, these methods have not become widespread. [116,[118][119][120][121][122][123][124][125][126][127][128][129] Tunable diode laser absorption spectroscopy (TDLAS) is an optical method that combines the traditional spectral method of measuring the absorption of a detectable substance with the use of tunable diode lasers as radiation sources.…”

“…Li et al [125] is devoted to the development of an optical measuring system for detecting hydrogen sulfide in the environment. The measurements are based on the WMS method using a single-mode distributed feedback laser with a central emission wavelength of 1.58 µm.…”

Feature Domain Transform Filter for the Removal of Inherent Noise Bound to the Absorption Signal

“…Moreover, the final effect of noise suppression is limited by the optimal selection of the number of window points and the fitting order. The empirical mode decomposition (EMD) algorithm can decompose the original signal into a series of intrinsic mode functions (IMFs). Compared with traditional time-frequency analysis techniques, EMD does not need to select the basis function, and its decomposition is based on the distribution of extreme points of the signal itself.…”

“…Moreover, the final effect of noise suppression is limited by the optimal selection of the number of window points and the fitting order. The empirical mode decomposition (EMD) 14 algorithm can decompose the original signal into a series of intrinsic mode functions (IMFs).…”