[A new linear neural network multi-component analysis method and its application in the analysis of VC yinqiao tablets quantitative analysis].

We measured NIR spectrum of VC yinqiao tablets with spectral instrument, analyzed the contents of acetaminophen and vitamin C in the VC yinqiao tablets with principal component analysis (PCA) and Linear Neural Network, and discussed the choice of principal component number and ANN's parameters affecting the network. To compare arithmetic performance, the authors also processed the spectral data with partial least squares and PCA-BP neural network. Compared with other two data process methods, the experiment and the result of data process showed that the PCA-linear neural network possess the best forecasting precision.

[Study on the method to adjust optical fiber's spectrum attenuation with the Fourier transform].

A method based on Fourier transform to compensate the non-linear attenuation of optical fiber used as a probe in a spectrum-collecting system was proposed. First the output electric currents of photoelectric tube with and without fiber were transformed to the frequency field. So an adjustable function in frequency field was obtained, and the adjustable function was transformed to the spectrum field, so the final adjustable function was obtained. A photoelectric system was designed for testing. With visible light, this method can make the error rate of fiber transmission as low as less than 1.54%. It is proved that the method is fit for adjusting some optical fiber spectrum attenuation.

[Improving partial least square regression precision in NIR multi-component analysis using artificial neural network].

The present paper presents a new NIR multi-component analysis method with Artificial Neural Network(ANN) and Partial Least Square Regression(PLS). First, this method divides the concentration range of training samples into some sub-ranges, and respectively computes a PLS correlation model in each sub-range with the sub-range's training samples. Then, the authors classify prediction samples according to its concentration sub-range with ANN and judge which sub-range theprediction sample belongs to. Finally, the authors compute the concentration of prediction component with the PLS correlation model of the sub-range according to ANN. The experiment and the result of data processing show that this method improves the model's applicability, and evidently enhances prediction precision compared to traditional PLS.

[The PLS analysis with short wave near-infrared of whole wheat for protein].

Short wave near-infrared spectrum of whole wheat was obtained by diffusion reflection. PLS method was used to analyze protein content of whole wheat. Different wavelength ranges were chosen for regression and information abstraction. The 3D curves were shown for different factors, prediction residual sum of squares (PRESS), and RMSECV. The best wavelength ranges and factors were determined. Analysis results for protein content of whole wheat were shown in three wavelength regions, and the predicted results were compared and discussed. The method of selecting advantageous wavelength ranges is feasible to obtain high prediction precision.

[Influence of smooth, 1st derivative and baseline correction on the near-infrared spectrum analysis with PLS].

This paper studied the influence of using pre-procession such as smooth, 1st derivative and baseline correction on the analysis of near-infrared spectrum. Comparing the analysis results by the pre-procession methods, and using PLS arithmetic, the best pre-procession was determined. In smooth pre-procession method, the best smooth points were proposed for regression using PLS. The analysis result is satisfactory.

[Research on model and wavelength selection of near infrared spectral information].

Based on stepwise linear regression, and according to the theory of near infrared absorbption, spectrum (1000-2500 nm) obtained by detector was divided into three ranges, which were I (1000-1400 nm) and II (1400-1860 nm) and III (1860-2500 nm). In each range the regression wavelengths of different wavelength gaps were picked up stepwise. Regression coefficients and parameters were calculated by Matlab5.3 Program. Regression models were built up in different ranges with different wavelength gaps. Best models could be determined. Prediction results of protein content of ground wheat were displayed in scatter plots. Different results were discussed and compared, which has referencemeaning for application.