A four-channel-based mid-infrared methane sensor system using novel optical/electrical dual-domain self-adaptive denoising algorithm

Abstract By incorporating two wide-band mid-infrared (MIR) incandescence light sources and two dual-channel pyroelectric detectors, a new four-channel-based mid-infrared gas sensor is presented for the measurement of methane (CH4) mole fractions. A least-square fast transverse filtering (LS-FTF) self-adaptive denoising algorithm was proposed to improve the robustness, stability, and noise immunity performance of the sensor. Numerical simulations were carried out to validate the function of the LS-FTF algorithm by introducing interference with different frequencies in the simulation. Sensor calibration and stability test were performed for the self-adaptive sensor using standard or diluted CH4 samples. With the intrinsic noise considered only, an Allan deviation of ∼ 207 parts per million by volume (ppmv) with a ∼ 3 s averaging time was obtained with no filter (NF) structure. Using self-adaptive filtering (SAF), the Allan deviation was decreased to ∼162 ppmv with the same averaging time. Noise with different modulation frequencies was imposed on the detection channel to validate the denoising performance of the four-channel structure. The reported four-channel-based sensor using LS-FTF technique shows an enhanced denoising performance as compared to the two or three-channel-based sensor structure.