摘要
为了研究中煤的发热量,采集了100个中煤样品的近红外漫反射光谱,采用主成分分析(PCA)对数据进行降维,建立定量的数学模型并与工业检测对比。分析结果表明,PC1的累计方差贡献率为92.13%,PC2的累计方差贡献率为91.35%;校正集和预测集的相关系数(R2)分别为0.961 54和0.880 64,校正集的均方根误差(RMSEC)和预测集的均方根误差(RMSEP)分别为0.173和0.300。实验结果表明:模型具有较高的相关性、稳定性和预测精度,为中煤发热量的近红外光谱定量检测奠定了基础。
To study the calorific value of middlings, collected the near-infrared diffuse reflection spectrum from 100 middling samples, principal components were extracted by principal component analysis(PCA), which established the mathematical model of quantitative and compared with industrial detection. PC1 accounts for 92.13% of cumulative variance contribution rate and PC2 accounts for91.35%; root mean square error of calibration(RMSEC) and root Mean Square Error of prediction(RMSEP)were 0.173 and 0.300, with correlation coefficients(R2) of 0.961 54 and 0.880 64, respectively. The results indicated that the model has a high relevance, stability and accuracy, which provided an approach for quantitative analysis of Calorific Value of middlings based on near infrared spectrum.
出处
《煤炭技术》
CAS
北大核心
2014年第6期218-220,共3页
Coal Technology
关键词
近红外光谱
主成分分析
中煤
发热量
near-infrared spectroscopy
PCA
middlings
calorific value
