摘要
对近红外光谱数据进行小波变换,利用处理后的小波系数,采用偏最小二乘法预测了有机肥料中钾离子(K+)的含量,建立了小波变换与近红外光谱技术结合用于测定奶牛粪便为主的有机堆肥产品样品中无机钾离子测定的模型。结果表明:小波变换充分提取了近红外光谱的信息,数据压缩为原始大小的3.6%,计算量大大减少;文章利用C4小波系数对48个有机肥料样本进行建模,对42个预示集样本进行预测,预示集的RMSEP(root mean square error of prediction)和r2(correlation coefficient)分别为0.113 8%和0.927,优于原始光谱直接建模的0.167 2%和0.835%。基于小波系数的模型优于传统的全谱模型,对于无机离子(K+)的测定可以取得较为准确的预测结果。
Potassium in organic fertilizer can be determined by near infrared(NIR) spectral technique,because the spectra are related to the organic groups with NIR absorption.A method for the determination of potassium iron(K+) in organic fertilizer samples was established based on the combination of discrete wavelet transform(DWT) and NIR technique.In the proposed method,the raw NIR data and their wavelet coefficients are used for modeling and prediction of the contents of potassium in organic fertilizer by partial least square(PLS) method.It is shown that there is almost no loss of spectral information with the NIR data compressed to 3.6% of its original size.The model based on wavelet coefficients is better than that based on the full NIR spectral range.With the improved method,accurate prediction can be achieved.
出处
《光谱学与光谱分析》
SCIE
EI
CAS
CSCD
北大核心
2007年第8期1523-1526,共4页
Spectroscopy and Spectral Analysis
基金
国家自然科学基金项目(20472076)
河南杰出人才创新基金项目(421002300)资助
关键词
近红外光谱
小波变换
偏最小二乘
有机堆肥
Near infrared spectrum
Wavelet transform
Partial least squares
Organic compost
