Abstract Gas sensors based on non-dispersive infrared (NDIR) technology have been researched extensively and dual-channel measurement schemes are widely adopted due to their unique advantages. However, there is a lack of comprehensive research on the performance of optical sensors with reference signals in existing two-channel measurement schemes. In this study, a methane gas detection system based on mid-infrared LED and photodiodes was designed with an empirical modal decomposition (EMD) algorithm for signal processing. The EMD processes measurement and reference signals to improve the signal-to-noise ratio (SNR) to 5 dB over the traditional white noise suppression method. The interference signal is identified and suppressed through joint analysis of the intrinsic mode function (IMFs) of the double-channel signals, thus further improving the SNR to 2 dB. The EMD is an adaptive method for processing non-stationary signals to optimize sensor performance.

中文翻译：

---

噪声抑制：非色散红外气体检测系统中的经验模态分解

摘要 基于非色散红外（NDIR）技术的气体传感器得到了广泛的研究，双通道测量方案因其独特的优势而被广泛采用。然而，现有的双通道测量方案中缺乏对具有参考信号的光学传感器性能的综合研究。在本研究中，设计了一种基于中红外 LED 和光电二极管的甲烷气体检测系统，采用经验模态分解 (EMD) 算法进行信号处理。EMD 处理测量和参考信号，以将信噪比 (SNR) 提高到传统白噪声抑制方法的 5 dB。通过对双通道信号的固有模态函数（IMFs）的联合分析，识别和抑制干扰信号，从而进一步将 SNR 提高到 2 dB。EMD 是一种用于处理非平稳信号以优化传感器性能的自适应方法。