摘要
近红外光谱数据量大 ,需要较大数据存储空间和较长的建模时间。本文以成品柴油性质分析为例 ,将小波变换用于近红外光谱数据压缩处理 ,详细考察了小波压缩参数 ,比较了压缩前后谱图差异以及性质分析偏差的变化。研究结果表明 ,采用Daubechies小波函数(N=2)为母函数 ,进行3次分解 ,直接采用其逼近系数(ca3)作为谱图压缩数据 ,其重构光谱与原始光谱基本一致。直接利用逼近系数进行性质分析 ,其分析精度与原始光谱数据基本相当 ,存储空间减少至原来的1/8 。
A method based on the wavelet transform(WT)was developed for the compression of the huge amount of data obtained from near infrared spectroscopy(NIR)for diesel.The compression variables for WT were studied in detail.The method was evaluated by comparing the spectra and analytical results obtained beˉfore and after the compression.The results showed that the NIR data was compressed to1/8of its original data,but the spectral information and analytical accuracy were not deteriorated.The compressed data can be directly used to establish the regression model with much shorter time.
出处
《分析测试学报》
CAS
CSCD
北大核心
2005年第1期17-20,24,共5页
Journal of Instrumental Analysis
关键词
近红外光谱
柴油
小波变换
数据压缩
Wavelet transform
Near infrared
Data compression
Diesel