中文摘要：

近红外光谱技术是光谱测试技术、化学计量学技术与计算机技术的有机结合，文章立足于食品有效成分无损定量检测的目标，介绍了微型近红外光谱分析系统的研制过程。作为系统测试的基础，文章重点研究了适用于在线实时分析的微型化近红外光谱仪，研制出的微型近红外光谱仪样机工作波长：850～1690nm，分辨率小于10nm。通过对样机主要参数的测试表明，性能接近国外同类产品水平。另外，化学计量学与计算机技术的结合是系统的核心，提出了基于最小二乘的局部加权回归（LS-LWR）的建模方法。最后，应用研制出的系统对葡萄糖水溶液定量测试实验表明，其预测集的相关系数达到了0．995，预测标准偏差为0．06。

英文摘要：

Near-infrared spectrometer is the integration of spectrum test technology, stoichiometry technology and computer technology. In the present paper, based on effective food ingredients and non-invasive quantitative detection, the development process of the micro-near-infrared spectrometer system was introduced. Spectrometer is the basis of the system. This paper focuses on the development of the micro-near-infrared spectrometer applicable to on-line real-time testing. A micro-near-infrared spectrometer prototype was developed successfully, its main technical parameter was tested, and the result shows： its operating wavelength is： 850-1 690 nm, optical resolution is： less than 10 um, and its performance has achieved the level of the congener foreign products. Stoichiometric technology and computer technology is the core of the system. LS-LWR modeling methods were proposed. Finally, the quantitative test for glucose water solution using the micro-near-infrared spectrometer shows that the correlation coefficient of prediction model is 0. 995, and the corresponding RMSEP is 0. 06.