小波分解重建算法提高近红外煤质水分检测精度  

Improving Precision in Coal Moisture Detection Using Wavelet Transform

作　　者：贾浩[1] 付强 韩婵娟[3] 邹德宝[1] 陈文亮[1,4] 徐可欣[1] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072 [2]天津市津南区环保监测站,天津300350 [3]天津大学环境科学与工程学院,天津300072 [4]天津大学微光机电系统技术教育部重点实验室,天津300072

出　　处：《光谱学与光谱分析》2012年第11期3010-3013,共4页Spectroscopy and Spectral Analysis

基　　金：国家自然科学重点基金项目(60938002);天津市自然科学基金重点项目(10JCZDJC22700)资助

摘　　要：水分是煤质经济价值的基本指标,会造成煤质利用率下降和能源浪费。因此,近红外光谱法实现煤质水分快速、无损、低成本检测具有重要意义。针对目前近红外煤质水分检测精度普遍在1%,且鲜有关于精度提高系统性研究的现状,从光谱预处理及波长选择进行比较,提出了提高煤质水分检测精度的可行性方法。实验结果表明,采用小波分解重建算法对1 300～2 400nm光谱区间进行降噪及去基线预处理,对重建光谱进行偏最小二乘建模,其建模结果明显优于其他方法,RMSEC达到0.06%,RMSEP达到0.27%,相关性系数R-square达到0.995,且证明模型稳定性较好,显著提高了近红外煤质水分的检测精度。Moisture,as a core determination of the economic value of coal,can result in the utilization and energy inefficiency.Near-infrared（NIR） spectroscopy,with advantages of high accuracy and low cost,provides significant solution to the quick and non-invasive detection of coal moisture.In the present paper,the improvement of the coal moisture analysis was conducted based on the precision of 1% and insufficient comparisons in recent experiments,and aspects of spectrum pretreatment and wavelength selection were mainly discussed.The optimized result with R-square of 0.995,RMSEC of 0.06% and RMSEP of 0.27% indicates the priority of wavelet decomposition and reconstruction,compared with other methods,in the noise reduction and baseline removing of original spectra（1 300-2 400 nm） before PLS modeling,and the stability experiment validates its robust potential in improving precision of coal moisture detection based on the NIR spectroscopy.

关 键 词：近红外光谱 小波分解重建 偏最小二乘法 

分 类 号：O657.3[理学—分析化学]