在现有近红外检测技术的基础上，设计了一种基于近红外的煤粉发热量检测系统。选取100个煤粉样品建立煤粉发热量的近红外模型，利用验证集的50个样本对模型的精度和稳定性进行了验证分析，结果表明：煤粉发热量的预测值与真实值的相关系数达到0995 8，相对偏差小于2%。系统具有良好的预测精度和稳定性，能够满足对煤粉发热量的快速检测的需求，而且该系统体积小、结构简单、操作方便，具有很好的可移植性。

On the basis of existing near infrared(NIR) detection technologies both at home and abroad, a pulverized coal calorific value detection system based on near infrared was designed. 100 pulverized coal samples were selected to create the near infrared model of pulverized coal colorific value, and then 50 samples of the validation set were used for validation analysis on the precision and stability of the model. It was turned out that the correlation coefficient between the forecast value and real value of pulverized coal colorific value was up to 0995 8, with relative deviation less than 2%. The result indicated that the system had a good forecast precision and stability, could meet the demands for rapid detection to the pulverized coal calorific value and the system was characterized by small volume, simple structure, convenient operation and had a very good transportability.