The Applicability Of The Near Infrared Spectrum Online Control Research

Nowaydays, it is widely recognized in aspects of quality control that a quality control during production areas is always better than the final detection. Therefore, the process analytical technology is becoming more and more important in production process, as well as the active ingredient detection process of traditional Chinese Medicine. Near-Infrared Spectroscopy has become the fastest growing process analysis technology for its advantages in aspects of quick-detection, easy-operation and green initiative in recent years. However, in the entire production process, it quite probably leads to the instability of the final products by those variations deriving from raw materials, intermediates or the equipment. Thus reliability evaluation of the process analytical method is necessary and essential. This study will mainly focus on the evaluation of the quantitative performance of the near-infrared spectrometer in order to provide a valid basis for the analysis of the active ingredients of traditional Chinese Medicine. This paper mainly includes the following sections.First, In order to guide the on-line detection of active components of traditional Chinese medicine, the on-line analysis method of NIRS was established for NH4Cl solution by using the near infrared spectroscopy technology. In addition, the impact of bubbles on the detection result was evaluated. The near-infrared spectroscopy technique was used to determine the content of NH4Cl. The prediction model was established using partial least squares (PLS) method after rejecting outliers. The predicted value calculated by the established model was quite close to the true value, which indicated a nice performance of the near infrared spectrometer in quantitative ability. The correlation coefficients (R2) and the root-mean-square error of cross-validation (RMSECV) of the quantitative calibration model were0.9910and0.4932respectively. The root-mean-square error of prediction (RMSEP) was0.2221. The relative prediction error (RESP) was less than10.00%when the concentration of NH4Cl was higher than1.5040mg·mL-1. In addition, the results showed that the impact of the bubble on the detection result was so small that it could be neglected. The error level of near-infrared spectroscopy technology could satisfy the on-line detection of traditional chinese medicine.Second, use a simple system of ammonium chloride to inspect the quantitative performance of the different instruments.(1) For the near-infrared Fourier spectrometer, took the ammonium chloride aqueous solution near-infrared absorption spectroscopy in the wavelength range of4000～10000cm-1, using Fourier transform near infrared spectrometer comes with data processing software TQ8.4for spectral preprocessing, this paper adopts modeling PLS method to establish quantitative models, and use the Mahalanobis distance to remove outliers, the spectrum were preprocessed by the first derivative method, the second derivative method, SG smoothing, Norris smoothing and compare these modeling results, select the first derivative of spectral pretreatment to eliminate the effect of the near-infrared spectrum of the scattered color and baseline drift.The RMSECV and correlation coefficient R2measure of the best modeling approach.Then use the model to predict the prediction set of samples. The model of the correlation coefficient R2was0.9994, RMSECV was0.0709. The NIR predicted actual results of the correlation coefficient was0.9900. The forecast results indicates good performance of the quantitative determination of near-infrared spectroscopy of a single sample of ammonium chloride.(2) Examine the quantitative performance of the FOSS instruments. FOSS instruments was grating near-infrared spectroscopy application CNC synchronization holographic grating optical system.500～2500nm scanning near infrared spectroscopy, use ammonium chloride to examine the quantitative performance of this instrument, use Vision software to preprocess spectra. Compare a variety of spectral pretreatment methods:ordinary guide, the second derivative,20point smoothing, baseline correction, multiple linear analysis, the SG smoothing single pretreatment as well as a variety of pretreatment combinations to get the best spectral pretreatment approach to building the best model. The results of analysis showed that the original spectrum can be good quantitative results, through the20point smoothing the baseplates line correction can get the best model, R2was0.9993, RMSECV was0.1097. Prediction set of validation results showed that the instrument has a good quantitative performance, and does not require complex spectral pretreatment can be obtained more reliable forecasts.(3) The investigation of quantitative performance portable acoustic-optic tunable filter (AOTF) inspection of the near-infrared spectrometer, scan ammonium chloride aqueous solution of the near infrared spectrum of this instrument in1100～2300nm band, through a series of spectra pretreatment methods to choose the best pretreatment methods, the use of ordinary guide+SG smoothing to eliminate noise and baseline drift for quantitative. Residual sum of squares method to eliminate outliers to optimization of the model. The model of the correlation coefficient was0.9937, RMSECV was0.1303. For verification results of the prediction set, the prediction correlation coefficient R2was0.9912, the instrument reliable and quantitative performance for its simplicity and portability, easy operation and are now widely used.Third, using near infrared spectroscopy for the rapid detection of plasticizer, using Fourier transform near infrared spectrometer scan of near-infrared spectroscopy in the10000 4000cm-1band range, through the different spectral preprocessing to remove noise and baseline drift the impact of the optimization model, excluding outliers using the Mahalanobis distance, which had a good prediction results, the near-infrared spectroscopy can be fast and accurate detection of plasticizer, which provide a fast and efficient method to detect beverage quality.