本文采用近红外漫反射光谱法对氢氧化铝粉末药品中主要成分氢氧化铝进行快速、无损定量分析。采用偏最小二乘法(PLS)建立近红外光谱信息与待测组分含量间的最佳数学校正模型。讨论了光谱的预处理方法和主成分数对PLS定量预测能力的影响,并对预测集样品含量进行预测,得到了较为满意的结果。 In this paper, near-infrared diffuse reflectance spectroscopy was used for rapid and non-destructive quantitative analysis of the main component of aluminum hydroxide in aluminum hydroxide powders. The least squares method (PLS) was used to establish the best mathematical model of calibration between the near-infrared spectral information and the content of the tested components. The effects of pretreatment methods and principal components of PLS ​​on the quantitative prediction ability of PLS ​​were discussed, and the prediction of PLS ​​contents was predicted. Satisfactory results were obtained.