近红外光谱煤质检测系统的设计与实现

Design and implementation of NIRS coal quality detection system

为降低检测成本及为高矿物样品无损检测技术提供保障,结合近红外检测技术原理,从硬件系统设计和软件系统设计两方面阐述了近红外光谱(NIRS)在线检测系统的设计与开发,采用面向对象的编程技术以实现在各种参数设置下的光谱采集、谱图显示、光谱预处理、校正模型的建立及未知样品的预测等功能,建立煤质水分含量的偏最小二乘定量模型,发现校正集和预测集的相关系数的平方分别为0.947和0.888,校正集均方根误差和预测集均方根误差分别为0.46和0.55,说明模型的预测能力较强,其应用前景良好。

In order to reduce the testing cost and provide technological guarantee for high mineral samples nondestructive examination,and combining with the technological principle of Near Infrared test,the design and development of online testing system of NIRS were demonstrated from two aspects（hardware system design and software system design）,object-oriented programming technology was used to realize the functions like spectrum acquisition,spectrogram display,the pretreatment of spectrum,establishment of calibration model and prediction of unknown samples under various parameter settings,the partial least squares quantitative model of coal quality moisture content was established,we found that the square of correlation coefficient of calibration set was 0.947 and correlation coefficient of prediction set was 0.888,the root-mean-square error of calibration set was 0.46 and the root-mean-square error of prediction set was 0.55,which meant that the prediction ability was relatively good,and the application prospect would be good.