【摘 要】
：
Spectroscopy presents a promising tool for process analysis and control.However,despite the identification capability of multivariate analysis,inherent low resolution and fluctuant spectral interferen
【机 构】
：
College of Precision Instrument and Opto-Electronics Engineering,Tianjin University,Tianjin,China
【出 处】
：
第二届国际过程分析与控制中国区论坛
论文部分内容阅读
Spectroscopy presents a promising tool for process analysis and control.However,despite the identification capability of multivariate analysis,inherent low resolution and fluctuant spectral interference often combine to limit conventional spectroscopy techniques [1].These disadvantages can limit the resolution of a spectral measurement,and may spoil the reliability of prediction [2].In this regard,we proposed a novel strategy,Multi-scale N-way partial least square(MN-PLS),with the systematic fusion of multi-scale regression and N-Way partial least square methods,to extract essential information from two dimensional correlation spectra(2D-COS)effectively [3].With MN-PLS,the pre-treatment process and multivariate calibration are integrated into a unified decision space,providing a great adaptability to process big spectral data with diversity,regardless of its structural characteristics.This would definitely promote the application of spectroscopy techniques in practice.Our work has tested and refined the MN-PLS strategy by drawing upon the quantitative analysis of dynamic multiple spectral data sets to yield challenges representative of those encountered in common spectral analysis of oil adulteration.This proposed strategy has improved the accuracy of olive adulteration discrimination significantly with refined division of two dimensional correlations time and frequency domains by weighting way.Satisfactory calibration results suggest the MN-PLS is promising to obtain reliable identification of olive adulteration with higher spectral resolution.
其他文献
A method of determination of multiple food colorants in carbonate beverages from ultraviolet-visible(UV-Vis)spectrum data using kernel independent component analysis(KICA)and support vector regression
A wide range of information is desired for optimizing chemical synthesis processes.Liquid chromatography is one tool that is useful for measuring the various components of the mixture.During the last
With the development of agricultural science and technology,the quality control of agricultural products was higher than before.On-line near infrared spectroscopy processing technology can be applied
为了对生产中的颗粒状原材料进行自动计数，提出一种基于计算机视觉的自动检测方法。该方法采用一种简单的拍照装置对颗粒物进行快速拍照，经过图像处理精确识别独立的颗粒，利用本研究提出的统计模型识别并估算粘连的颗粒，最后得到精确的计数。使用人工计数的方式对结果进行检测，根据修正的结果重新训练模型，通过机器学习的方法不断提升模型的精度。实际应用结果表明，该方法具有很高的精度可很好地解决颗粒物粘连的问题。
At present,tradition Chinese medicine(TCM)industry in China is in the stage from the empirical development to engineered production.Near infrared spectroscopy(NIRS),as a kind of process analysis techn
On-line mass spectrometry(MS)has been developed and implemented as a tool to monitor and quantify a number of PAT & QbD applications.These include monitoring and control of fermentors,bioreactors and
Process understanding is one of the key elements in todays pharmaceutical industry thanks to FDAs "Pharmaceutical cGMPs for the 21st Century: A Risk-Based Approach." However,for many generic pharmaceu
For pharmaceutical powders,it has been well studied and accepted that effective penetration depth of NIR beam is about 1 mm or less and more than 90%of the information is obtained from a thin layer of
Cashmere is a rare special animal fiber in our country.It is very difficult to be identified because of its similar chemical properties with wool.Using microscope to identify cashmere is currently one