摘要
利用小波变换的时频局部变换特性对近红外光谱数据进行滤噪处理,能有效地从信号中提取有用信息,从而提高信号的信噪比,提高模型的预测精度和稳健性.将该方法应用于近红外血糖浓度测量的基础研究中,实验结果表明,校正模型的预测标准偏差(RMSEP)分别减少了53%和58%.这对于血糖浓度测量的研究具有一定的现实意义.
Because of its properties of time-frequency transform, the wavelet transform is an effective denoised method. With the pretreatment of spectra based on wavelet analysis, the spectra-to-noise ratio was greatly improved while the noise was suppressed effectively. It also improved the prediction precision and robust of the model. This method is applied to fundamental study of measurement of blood glucose concentration with spectroscopy. Experimental results show that the Root mean square error of prediction (RMSEP) of the calibration model reduces by 53% and 58% respectively. It is instructive for the study of the theory of measurement of blood glucose with spectroscopy.
出处
《天津大学学报（自然科学与工程技术版）》
EI
CAS
CSCD
北大核心
2004年第6期535-539,共5页
Journal of Tianjin University：Science and Technology
基金
国家自然科学基金资助项目(30170261)
国家"十五"攻关项目(BA706B 21)
教育部科学技术研究重点项目
南开大学 天津大学刘徽应用数学中心资助项目.
关键词
小波变换
人体血糖浓度
稳健性
wavelet transform
human blood glucose concentration
robust
