Abstract The purpose of this work was to develop a new multivariate calibration method using partial least squares (PLS) regression to quantify methane (CH4) in samples of gases eructed by ruminants and exhaled together with the respiration of cows. Four Holstein cows were confined and received a diet of 18.7 kg of dry matter per day for 4 days. During this period, samples of the gas exhaled by these cows were collected. Gas sampling was performed using evacuated PVC canisters during a period of 24 h, and in this case, two gas samples were collected per animal as well as two samples of the background environment in which the animals were confined. In total, 40 samples were collected and submitted to analysis. Samples were analyzed by gas chromatography and then by infrared spectroscopy using a long optical path gas cell. The results of both analyses were used to develop the multivariate calibration model, PLS regression. The results show a high correlation between chromatography method and infrared analyses with values of determination coefficients greater than 0.96 and root mean square error (RMSE) values lower than 23.3 ppm.

中文翻译：

---

用红外光谱法和多元方法量化奶牛排放的气体样品中的甲烷

摘要 这项工作的目的是开发一种新的多元校准方法，使用偏最小二乘法 (PLS) 回归来量化反刍动物喷出的气体样本中的甲烷 (CH4)，并与奶牛的呼吸一起呼出。四头荷斯坦奶牛被限制并接受每天 18.7 公斤干物质的饮食，持续 4 天。在此期间，收集了这些奶牛呼出的气体样本。在 24 小时内使用抽真空的 PVC 罐进行气体采样，在这种情况下，每只动物收集两个气体样本以及动物被限制在其中的背景环境的两个样本。总共收集了 40 个样品并提交进行分析。样品通过气相色谱分析，然后使用长光程气室通过红外光谱分析。两种分析的结果都用于开发多变量校准模型，PLS 回归。结果表明，色谱法与红外分析之间存在高度相关性，测定系数值大于 0.96，均方根误差 (RMSE) 值小于 23.3 ppm。