基于天然气组分红外光谱图的数据预处理方法研究

利用红外光谱分析仪对天然气组分进行组分分析时所获得光谱信号往往会受杂散光、噪声、基线漂移等因素的干扰，从而影响最终定量分析结果，故需要在建模前对原始光谱进行预处理。为解决仪器测量光谱图的噪声干扰问题，本文提出一种Savitzky-Golay平滑法结合sym6小波函数软阈值去噪法对光谱图进行预处理。将传统的预处理方法与SG平滑法结合小波函数法进行对比分析。结果表明，以SG平滑法结合sym6小波函数软阈值去噪法对光谱图进行预处理，其拟合优度数值最高为0.98652，残差平方和数值最低为5.50694，证明使用该方法后的函数分峰拟合效果最佳，处理效果优于传统方法。

Data Preprocessing Method for Infrared Spectra Analysis of Natural Gas Components

When using infrared spectroscopy to analyze the components of natural gas, the obtained spectral signals often contain interference from stray light, noise, baseline drift, and other factors, which affects the resulting quantitative analysis. Therefore, it is necessary to preprocess the original spectrum before modeling. As a potential solution, an SG smoothing method combined with the soft threshold denoising method of the sym6 wavelet function was proposed to preprocess the spectrogram. The traditional preprocessing method and the proposed method are compared and analyzed. The results show that when the proposed method is used to preprocess the spectrogram, the highest goodness of fit value is 0.98652, and the lowest residual sum of squares value is 5.50694, which proves that the function peak fitting effect is the best after using this method, and the processing effect is better than that of the traditional method.