【作者】 包春芳；

【导师】 任玉林；

【作者基本信息】 吉林大学 ， 分析化学， 2009， 硕士

【摘要】 食品安全问题已经成为人们日益关注的焦点,传统的分析方法操作复杂,样品需要预处理,分析时间长且检测成本高,发展适合食品的过程分析方法,实现生产过程的质量监测和优化控制,保证食品质量的稳定性和均一性是急需解决的难题之一。本文针对酱油、红酒、黄酒等液态调味品、酒类组成复杂、检测方法繁琐的问题,将近红外光谱技术与偏最小二乘法(PLS)结合,分别对三种液态食品的主要质量指标进行了检测分析,实现了液态食品的无损非破坏质量分析。建立了近红外光谱技术与偏最小二乘法(PLS)结合对液态食品非破坏质量分析的模型。偏最小二乘法(PLS)是目前近红外光谱分析中应用最广的化学计量学方法之一,它能有效地除去噪声,解决光谱共线问题,因此它能够建立更加精确的数学校正模型。本文研究用PLS建立酱油、红酒、黄酒近红外光谱校正模型,同时讨论了光谱预处理方法和主成分数对PLS模型预测精度的影响。并对相关质量指标进行定量预报,结果令人满意,实验结果具有较高的预报精度。本文对近红外光谱技术在液态食品非破坏质量分析技术的研究,有助于提高该技术在食品安全领域的理论研究和应用,有利于解决目前食品质量控制中缺乏快速分析方法的问题。研究结果表明,本文所获得的结论,可推广到食品快速质量分析和食品质量的在线检测中,具有较强的创新意义和实际应用价值。更多还原

【Abstract】 In recent years, near infrared (NIR) spectroscopy has been spread rapidly as a powerful analytical tool. NIR spectroscopy analysis technique is simple, convenient, fast, non-destructive and no chemical contamination; it has been widely used in agric- ulture, food, petrochemical, textile and pharmaceutical industries etc. The NIR spectrum refers to the electromagnetic spectrum range from 780 nm to 2500 nm. Normally,NIR region is divided into two regions,spectrum from 780 nm to 1100 nm called Short Wavelength Near-infrared and spectrum from 1100 nm to 2500 nm called Long Wavelength Near-infrared.The NIR spectrum is the absorption generated by frequency doubling and sum- frequency correspond to the vibrations of O-H, C-H and N-H groups in the compound. However, because the absorption of NIR spectra is weak, seriously overlapping and lack of characteristic peak, it is hard to analyze the data just in traditional ways, and some data processing methods must be applied in order to obtain the exact result in the qualitative and quantitative analysis. Therefore, the modern NIR technique is formed by the combination of NIR spectroscopy and chemometrics methodology. Combined with the strong information process ability of chemometrics, NIR spectroscopy can fully demonstrate its rapidness, precision and non-destructive analysis which has great potential in these application. The linear regression modeling methods that are commonly used in chemometrics including multiple linear regression (MLR), principal component regression (PCR) and partial least squares (PLS). Among these methods, PLS has been used more and more widely as a standard method, due to its simple calculation, high quality calibration models and large amo- unt of business software. Through the establishment of optimal calibration models, accurate predication result can be obtained.With the improvement of our living standard and the popularization of food and nutrition hygiene knowledge, food safety problem has been increasingly becoming a focus. The conventional quality control method need to destroy samples, which is tedious and of high cost, so fake food can not be detected in time,which brings great harm to our life and safety. Therefore, it is urgent to develop a food analysis method to inspect the producing process, optimize the quality control, and ensure the stability and homogeneity of food. In this paper, a method for nondestructive, expeditious determination of the main quality indexes of soy sauce, red wine and yellow wine is investigated by the use of near-infrared spectroscopy combined with partial least squares (PLS) method. Experimental results demonstrate that the optimal model designed with this method has a high ability for prediction, thus it is suitable for the non-destructive quality control of food.The main research contents of this dissertation are described as follows:1. Different spectral preprocessing methods were studied. Original spectra include information concerning sample composition, as well as noises from various aspects. The analyte of interest absorbs can be disturbed by these noisy signals. Therefore, the original spectra must be pretreated. The general preprocessing methods commonly used including first-derivation and second-derivation. The calibration models design- ed with the original and different preprocessing methods were studied and discussed, and the result shows that the pretreatment of the original spectra is necessary. Through the application of these methods, the noise of NIR spectra can be removed or reduced, which set up solid foundation for the further establishment.2. The method of near-infrared spectroscopy combined with partial least squares (PLS) was applied to the nondestructive qualitative determination of liquid food. Owing to the complex compositions of liquid condiment and alcoholic drink, the obta- ined NIR spectra were serious overlapped, which results in difficulties in qualitative analysis, but that could be solved by the combination of NIR spectroscopy with chem- ometrics method. At present, PLS is one of the widely used chemometrics method in NIR analysis, this method can eliminate the noises effectively and solves the collinear problem of spectra, establishing the optimal mathematic calibration models.In this paper, the scores and optimal PLS factors that affect the predictive ability of PLS models were discussed, and the optimal calibration models were obtained. Furth- ermore, the main quality indexes were determined, the results were satisfied, and it has a high ability for prediction. Experimental results show that the optimal models designed with PLS method were fast and convenient, thus it is feasible for the application.Through the study of liquid condiment and alcoholic drink, the models were established by near-infrared spectroscopy combined with partial least squares (PLS) method, and accurate result was obtained. In the experiment, via the improvement of method and modification of model parameters, the research of food analysis model has achieved the purpose of systematic and scientific. The experimental results demo- nstrate that NIR spectroscopy combined with PLS method was practicable in the food qualitative analysis.Therefore, the study on nondestructive qualitative analysis of liquid food with NIR spectroscopy can be extended to the expeditious analysis and on-line inspection of food, and the research will have a broad exploration prospect and practical application value in the field of food analysis.更多还原