植物油的红外光谱结合神经网络快速识别

为实现对植物油的快速检测,借助衰减全反射-傅里叶变换红外光谱分析技术并结合深度学习算法对植物油开展光谱模式识别工作。实验获取8种植物油样本的光谱数据,采用标准正态变换和一阶导数预处理方法消除背景干扰,同时采用竞争性自适应重加权算法模型对各样本特征光谱数据进行提取,分别建立长短记忆神经网络(LSTM)、基于Levenberg-Marquardt算法改进的BP神经网络对提取特征波长后的植物油种类进行预测识别与比较,并采用后者进行了实际样品的识别检测。结果表明,通过提取特征波长,可有效提高LSTM模型的识别准确率,其最优准确率从提取特征波长前的30%～40%提高到80%～90%,模型运行时间从提取特征波长前的111 min 25 s缩短至1 min 45 s。相较于LSTM模型,基于Levenberg-Marquardt算法改进的BP神经网络的分类识别准确率更高,达到99.852%,用于实际样品的识别,识别准确率达到100%。实验结果可为植物油的无损快速检验提供一定的参考与借鉴。     

Approaches to Adulteration Detection in Instant Coffees using Infrared Spectroscopy and Chemometrics

Fourier transform infrared (FTIR) spectroscopy is examined as a rapid alternative to wet chemistry methods for the detection of adulteration of freeze-dried instant coffees. Spectra have been collected of pure coffees, and of samples adulterated with glucose, starch or chicory in the range 20–100 g kg⁻¹. Two different FTIR sampling methods have been employed: diffuse reflectance, and attenuated total reflectance. Three different statistical treatments of the spectra were carried out. Firstly, the spectra were compressed by principal component analysis and a linear discriminant analysis performed. With this approach, a 98% successful classification rate was achieved. Secondly, a simultaneous partial least square regression was carried out for the content of three added carbohydrates (xylose, glucose and fructose) in order to assess the potential of FTIR spectroscopy for determining the carbohydrate profile of instant coffee. Lastly, the discrimination of pure from adulterated coffee was performed using an artificial neural network (ANN). A perfect rate of assignment was obtained. The generalization ability of the ANN was tested on an independent validation data set; again, 100% correct classifications were achieved.     

基于近红外光谱与深度学习的紫菜水分无损检测

为探索近红外光谱结合深度学习网络对紫菜水分定量检测的可行性,本研究检测并收集了479组干条斑紫菜的光谱数据和水分含量数据,分别使用四种方法对其中的光谱数据进行了预处理,并在全波段下建立了四种传统定量水分预测模型和一种卷积神经网络（Convolution Neural Networks,CNN）深度学习水分预测模型。对比五种模型预测结果后发现,在S-G平滑结合二阶导数的预处理方法下所建立的CNN模型预测效果最佳,其预测均方根误差（Root-Mean-Square Error of Prediction,RMSEP）值为0.456,预测集决定系数（Coefficient of Determination of Prediction,R_p²）值为0.990,优化后,该模型的RMSEP值降至0.342,R_p²值可以达到0.994（>0.8）,同时,外部验证相对误差（Ratio of Performance to Deviation for Validation,RPD）值达6.155（>3）,证明了模型实际应用于农业和食品工业的可能性。该CNN模型能够快速、准确、无损地预测条斑紫菜的水分含量,提高了紫菜水分检测的效率和准确性,为相关干制水产品的质量控制提供了重要的参考依据。     

The use of Fourier-transform infrared spectroscopy to predict cheese yield and nutrient recovery or whey loss traits from unprocessed bovine milk samples

Cheese yield is an important technological trait in the dairy industry in many countries. The aim of this study was to evaluate the effectiveness of Fourier-transform infrared (FTIR) spectral analysis of fresh unprocessed milk samples for predicting cheese yield and nutrient recovery traits. A total of 1,264 model cheeses were obtained from 1,500-mL milk samples collected from individual Brown Swiss cows. Individual measurements of 7 new cheese yield-related traits were obtained from the laboratory cheese-making procedure, including the fresh cheese yield, total solid cheese yield, and the water retained in curd, all as a percentage of the processed milk, and nutrient recovery (fat, protein, total solids, and energy) in the curd as a percentage of the same nutrient contained in the milk. All individual milk samples were analyzed using a MilkoScan FT6000 over the spectral range from 5,000 to 900 wavenumber × cm⁻¹. Two spectral acquisitions were carried out for each sample and the results were averaged before data analysis. Different chemometric models were fitted and compared with the aim of improving the accuracy of the calibration equations for predicting these traits. The most accurate predictions were obtained for total solid cheese yield and fresh cheese yield, which exhibited coefficients of determination between the predicted and measured values in cross-validation (1-VR) of 0.95 and 0.83, respectively. A less favorable result was obtained for water retained in curd (1-VR = 0.65). Promising results were obtained for recovered protein (1-VR = 0.81), total solids (1-VR = 0.86), and energy (1-VR = 0.76), whereas recovered fat exhibited a low accuracy (1-VR = 0.41). As FTIR spectroscopy is a rapid, cheap, high-throughput technique that is already used to collect standard milk recording data, these FTIR calibrations for cheese yield and nutrient recovery highlight additional potential applications of the technique in the dairy industry, especially for monitoring cheese-making processes and milk payment systems. In addition, the prediction models can be used to provide breeding organizations with information on new phenotypes for cheese yield and milk nutrient recovery, potentially allowing these traits to be enhanced through selection.     

Cheddar cheese classification based on flavor quality using a novel extraction method and Fourier transform infrared spectroscopy

Analysis of Cheddar cheese flavor using trained sensory and grading panels is expensive and time consuming. A rapid and simple solvent extraction procedure in combination with Fourier transform infrared spectroscopy was developed for classifying Cheddar cheese based on flavor quality. Fifteen Cheddar cheese samples from 2 commercial production plants were ground into powders using liquid nitrogen. The water-soluble compounds from the cheese powder, without interfering compounds such as fat and protein, were extracted using water, chloroform, and ethanol. Aliquots (10 μL) of the extract were placed on a zinc selenide crystal, vacuum dried, and scanned in the mid-infrared region (4,000 to 700 cm⁻¹). The infrared spectra were analyzed by soft independent modeling of class analogy (SIMCA) for pattern recognition. Sensory flavor quality of these cheeses was determined by trained quality assurance personnel in the production facilities. The SIMCA models provided 3-dimensional classification plots in which all the 15 cheese samples formed well-separated clusters. The orientation of the clusters in 3-dimensional space correlated well with their cheese flavor characteristics (fermented, unclean, low flavor, sour, good Cheddar, and so on). The discrimination of the samples in the SIMCA plot was mainly due to organic acids, fatty acids and their esters, and amino acids (1,450 to 1,350 and 1,200 to 990 cm⁻¹), which are known to contribute significantly to cheese flavor. The total analysis time, including the sample preparation time, was less than 20 min per sample. This technique can be a rapid, inexpensive, and simple tool to the cheese industry for predicting the flavor quality of cheese.     

A comparison and joint use of VIS-NIR and MIR spectroscopic methods for the determination of some chemical parameters in soft cheeses at external and central zones: a preliminary study

There is a strong tendency towards exploring rapid and low cost methods for determining chemical parameters and degree of the ripening of cheeses. The visible-near infrared (VIS-NIR), mid infrared (MIR) and combination of VIS-NIR and MIR spectroscopic methods for measurements of some selected parameters of soft cheeses were compared. Fifteen traditional and stabilised retail soft cheeses, differing in manufacturing process were studied. Fat, dry matter (DM), pH, total nitrogen (TN) and water soluble nitrogen (WSN) contents were determined by reference methods and scanned with VIS-NIR and MIR spectrophotometers in reflectance mode. Three separate prediction models were developed from the VIS-NIR, MIR and the joint VIS-NIR-MIR spectra using the partial least square (PLS) regression and leave one-out cross-validation technique. Results showed that fat, DM, TN and WSN were the best predicted with the VIS-NIR models providing the lowest values of the root mean square error of prediction (RMSEP) of 1.32, 0.70, 0.11 and 0.10, respectively. The combination of the VIS-NIR and MIR spectral improved slightly the prediction of only the pH. This suggests using the VIS-NIR for the determination of fat, DM, TN and WSN. The pH can also be predicted from the two techniques with approximate quantitative prediction, while a difference between low and high levels of WSN/TN ratio could be determined by the VIS-NIR, MIR or joint use of VIS-NIR-MIR.     

Evaluation of salt whey as an ingredient in processed cheese

The objective of this research was to determine whether salt whey, obtained from a traditional Cheddar cheese manufacturing process, could be used as an ingredient in processed cheese. Due to its high salinity level, salt whey is underutilized and leads to disposal costs. Consequently, alternative uses need to be pursued. The major components of salt whey (salt and water) are used as ingredients in processed cheese. Three replicates of pasteurized processed cheese (PC), pasteurized processed cheese food (PCF), and pasteurized processed cheese spread (PCS) were manufactured. Additionally, within each type of processed cheese, a control formula (CF) and a salt whey formula (SW) were produced. For SW, the salt and water in the CF were replaced with salt whey. The composition, functionality, and sensory properties of the CF and SW treatments were compared within each type of processed cheese. Mean melt diameter obtained for the CF and SW processed cheeses were 48.5 and 49.4 mm, respectively, for PC, and they were 61.6 and 63 mm, respectively, for PCF. Tube-melt results for PCS was 75.1 and 79.8 mm for CF and SW treatments, respectively. The mean texture profile analysis (TPA) hardness values obtained, respectively, for the CF and SW treatments were 126 N and 115 N for PC, 62 N and 60 N for PCF, and 12 N and 12 N for PCS. There were no significant differences in composition or functionality between the CF and SW within each variety of processed cheese. Consequently, salt whey can be used as an ingredient in PC without adversely affecting processed cheese quality.     

近红外光谱结合极限学习机的榛子蛋白质含量检测

以毛榛与平榛作为研究对象,将去壳后的毛榛与平榛分别制成90组与60组试验样本,获取波长范围为900 nm～1700 nm的原始光,同时,通过凯氏定氮法测得其中蛋白质含量真实值,并通过对比研究一阶(1-der)导数、二阶(2-der)导数、多元散射校正(multiplicative scatter correction,...     

Comparison between genetic parameters of cheese yield and nutrient recovery or whey loss traits measured from individual model cheese-making methods or predicted from unprocessed bovine milk samples using Fourier-transform infrared spectroscopy

Cheese yield is an important technological trait in the dairy industry. The aim of this study was to infer the genetic parameters of some cheese yield-related traits predicted using Fourier-transform infrared (FTIR) spectral analysis and compare the results with those obtained using an individual model cheese-producing procedure. A total of 1,264 model cheeses were produced using 1,500-mL milk samples collected from individual Brown Swiss cows, and individual measurements were taken for 10 traits: 3 cheese yield traits (fresh curd, curd total solids, and curd water as a percent of the weight of the processed milk), 4 milk nutrient recovery traits (fat, protein, total solids, and energy of the curd as a percent of the same nutrient in the processed milk), and 3 daily cheese production traits per cow (fresh curd, total solids, and water weight of the curd). Each unprocessed milk sample was analyzed using a MilkoScan FT6000 (Foss, Hillerød, Denmark) over the spectral range, from 5,000 to 900 wavenumber × cm⁻¹. The FTIR spectrum-based prediction models for the previously mentioned traits were developed using modified partial least-square regression. Cross-validation of the whole data set yielded coefficients of determination between the predicted and measured values in cross-validation of 0.65 to 0.95 for all traits, except for the recovery of fat (0.41). A 3-fold external validation was also used, in which the available data were partitioned into 2 subsets: a training set (one-third of the herds) and a testing set (two-thirds). The training set was used to develop calibration equations, whereas the testing subsets were used for external validation of the calibration equations and to estimate the heritabilities and genetic correlations of the measured and FTIR-predicted phenotypes. The coefficients of determination between the predicted and measured values in cross-validation results obtained from the training sets were very similar to those obtained from the whole data set, but the coefficient of determination of validation values for the external validation sets were much lower for all traits (0.30 to 0.73), and particularly for fat recovery (0.05 to 0.18), for the training sets compared with the full data set. For each testing subset, the (co)variance components for the measured and FTIR-predicted phenotypes were estimated using bivariate Bayesian analyses and linear models. The intraherd heritabilities for the predicted traits obtained from our internal cross-validation using the whole data set ranged from 0.085 for daily yield of curd solids to 0.576 for protein recovery, and were similar to those obtained from the measured traits (0.079 to 0.586, respectively). The heritabilities estimated from the testing data set used for external validation were more variable but similar (on average) to the corresponding values obtained from the whole data set. Moreover, the genetic correlations between the predicted and measured traits were high in general (0.791 to 0.996), and they were always higher than the corresponding phenotypic correlations (0.383 to 0.995), especially for the external validation subset. In conclusion, we herein report that application of the cross-validation technique to the whole data set tended to overestimate the predictive ability of FTIR spectra, give more precise phenotypic predictions than the calibrations obtained using smaller data sets, and yield genetic correlations similar to those obtained from the measured traits. Collectively, our findings indicate that FTIR predictions have the potential to be used as indicator traits for the rapid and inexpensive selection of dairy populations for improvement of cheese yield, milk nutrient recovery in curd, and daily cheese production per cow.     

木瓜凝乳蛋白酶在干酪及其副产物乳清中的应用

随着干酪产业的发展，凝乳酶替代品的研究引起了人们的广泛关注。综述了凝乳酶替代品的研究现状及木瓜凝乳蛋白酶的研究进展，并展望了木瓜凝乳蛋白酶在乳品生产中的应用前景。     

回归神经网络红外光谱法同时测定丙酮中苯、甲苯的含量

采用一种基于回归神经网络模型的新方法，对丙酮中所含红外光谱相互重叠的苯及甲苯同时进行定量测定，与最小二乘法处理所得结果相对照，此法优于最小二乘法     

Invited review: Use of infrared technologies for the assessment of dairy products—Applications and perspectives

Dairy products are important sources of nutrients for human health and in recent years their consumption has increased worldwide. Therefore, the food industry is interested in applying analytical technologies that are more rapid and cost-effective than traditional laboratory analyses. Infrared spectroscopy accomplishes both criteria, making real-time determination feasible. However, it is crucial to ensure that prediction models are accurate before their implementation in the dairy industry. In the last 5 yr, several papers have investigated the feasibility of mid- and near-infrared spectroscopy to determine chemical composition and authenticity of dairy products. Most studies have dealt with cheese, and few with yogurt, butter, and milk powder. Also, the use of near-infrared (in reflectance or transmittance mode) has been more prevalent than mid-infrared spectroscopy. This review summarizes recent studies on infrared spectroscopy in dairy products focusing on difficult to determine chemical components such as fatty acids, minerals, and volatile compounds, as well as sensory attributes and ripening time. Promising equations have been developed despite the low concentration or the absence of specific absorption bands (or both) for these compounds.     

Classification of dry-cured hams according to the maturation time using near infrared spectra and artificial neural networks

An attempt to classify dry-cured hams according to the maturation time on the basis of near infrared (NIR) spectra was studied. The study comprised 128 samples of biceps femoris (BF) muscle from dry-cured hams matured for 10 (n = 32), 12 (n = 32), 14 (n = 32) or 16 months (n = 32). Samples were minced and scanned in the wavelength range from 400 to 2500 nm using spectrometer NIR System model 6500 (Silver Spring, MD, USA). Spectral data were used for i) splitting of samples into the training and test set using 2D Kohonen artificial neural networks (ANN) and for ii) construction of classification models using counter-propagation ANN (CP-ANN). Different models were tested, and the one selected was based on the lowest percentage of misclassified test samples (external validation). Overall correctness of the classification was 79.7%, which demonstrates practical relevance of using NIR spectroscopy and ANN for dry-cured ham processing control.     

红外光谱快速测定油茶籽油脂肪酸组成的模型建立

通过气相色谱和傅里叶红外光谱仪测定86个油茶籽油样本的脂肪酸组成和红外光谱图,采用支持向量机(SVM)和BP人工神经网络(ANN)的非线性建模方法,构建油茶籽油中主要脂肪酸的定量回归模型。结果表明:ANN建立的油酸和棕榈酸定量回归模型精确度比SVM高,校正集的相关系数(R)分别为0.9987和0.9451,预测集的相关系数分别为0.9557和0.9262,相对标准偏差分别小于1%和5%;SVM和ANN建立的亚油酸定量分析模型精确度都非常高,相对标准偏差均小于1%。说明红外光谱用于油茶籽油中主要脂肪酸的快速检测是完全可行的。     

超滤法浓缩乳清蛋白前处理研究

为了提高乳清在进行超滤的通透性,从而提高超滤的效率,对利用CaCl2絮凝沉淀乳清的工艺条件进行研究,通过单因素和正交试验优化,综合考虑蛋白含量、脂肪含量和透光率3个指标,确定最佳的絮凝工艺条件为pH7.5、1mol/L CaCl2溶液添加量12mL/L、在60℃水浴缓慢搅拌絮凝15min。在最佳絮凝工艺条件下进行验证,结果表明：经过预处理后的乳清蛋白损失率只有13.6%,脂肪去除率达到86%以上,透光率也有显著提高。     

Near-Infrared Reflectance Spectroscopy for Rapid Analysis of Curds during Cheddar Cheese Making

Near-infrared reflectance spectroscopy (NIRS) equations were developed for rapid analysis of curds during Cheddar cheese making. The coefficients of determination (R²) for NIRS and chemical analysis were: moisture (0.982), protein (0.965), fat (0.951), and lactose (0.909). When validation samples were compared by NIRS and chemical analysis, the R²s were moisture (0.984), protein (0.964), fat (0.957), and lactose (0.982). These results suggest that NIRS is applicable for rapidly monitoring chemical changes in curds.     

基于机器学习对食品安全的调控与分析

民以食为天,食以安为先。食品质量及其安全性关系到国计民生。随着中国经济的发展和人民生活质量的提高,食品行业的规模也逐年壮大,社会和消费者对食品生产的质量及其本身安全性有了更加严格的要求。但是,食品质量安全事件时有发生,使得食品质量安全的管理成为了改善民生的重要任务。机器学习已在食品质量与安全领域被广泛应用,它具有自主学习能力强、非线性拟合能力好、建模快速等特点,其中的神经网络模型和监督学习方法能够准确、快速的对食品在生产过程中进行质量检测与过程控制。本文将着重阐述机器学习在食品质量与安全领域中的研究进展,以食品质量检验、食品过程追溯、食品安全预警3个方向进行论述。以期阐明机器学习算法在食品调控环节中的侧重点、优缺点和未来发展方向,为保障食品质量与安全的智能化发展提供理论支持与技术指导。     

Production of a high gel strength whey protein concentrate from cheese whey

In order to develop a process for the production of a whey protein concentrate (WPC) with high gel strength and water-holding capacity from cheese whey, we analyzed 10 commercially available WPC with different functional properties. Protein composition and modification were analyzed using electrophoresis, HPLC, and mass spectrometry. The analyses of the WPC revealed that the factors closely associated with gel strength and water-holding capacity were solubility and composition of the protein and the ionic environment. To maintain whey protein solubility, it is necessary to minimize heat exposure of the whey during pretreatment and processing. The presence of the caseinomacropeptide (CMP) in the WPC was found to be detrimental to gel strength and water-holding capacity. All of the commercial WPC that produced high-strength gels exhibited ionic compositions that were consistent with acidic processing to remove divalent cations with subsequent neutralization with sodium hydroxide. We have shown that ultrafiltration/diafiltration of cheese whey, adjusted to pH 2.5, through a membrane with a nominal molecular weight cut-off of 30,000 at 15 degrees C substantially reduced the level of CMP, lactose, and minerals in the whey with retention of the whey proteins. The resulting WPC formed from this process was suitable for the inclusion of sodium polyphosphate to produce superior functional properties in terms of gelation and water-holding capacity.