| Long shelf life yogurt has a rich taste and high nutritional value,as lactose is extensively decomposed during its fermentation process,which is of great significance for the large population of lactose intolerant people in Asia.However,as a new product,research is still in its infancy,and comprehensive research on changes in its physicochemical indicators and shelf life prediction models is in a blank state,requiring further research.Ultra high temperature sterilized milk is a product with a wide range of circulation on the market,characterized by high nutritional value and rich taste.It accounts for about 80% of the market share in China’s dairy product market and meets the drinking milk needs of most Chinese people.However,due to the complex composition of ultra high temperature sterilized milk and the interaction between nutrients,it has always been difficult to achieve accurate prediction for products with different formulas.This article takes room temperature yogurt and UHT milk as research objects,and considers the influence of temperature to establish shelf life prediction models for solidified and stirred room temperature yogurt;Based on chemical kinetics,multilayer perceptron neural network and back-propagation neural network,a multi factor UHT milk packaging shelf life prediction model was constructed by integrating initial fat content,initial protein content and storage temperature.The main research content and achievements are as follows:(1)Establishment of a quality change and packaging shelf life prediction model for room temperature yogurt during storage.Using a single factor experiment,two types of stirred yogurt and solidified yogurt were stored at temperatures of 25,35,and 45 ℃ to study the effect of temperature on the sensory quality and physicochemical index changes of room temperature yogurt storage.Based on the sensory scores and physicochemical index data of room temperature yogurt,correlation and principal component analysis were conducted to determine that browning index and water holding capacity were the key shelf life indicators of stirred yogurt and solidified yogurt,respectively,The chemical kinetics fitting was carried out respectively,and two more reliable packaging shelf life prediction models of the characteristics of long shelf life yogurt were obtained.(2)Determination of quality changes and characteristic quality indicators during UHT milk storage.Experimental determination and analysis of the changes in physicochemical properties of UHT milk samples during storage at 23 ℃ and 37 ℃.The results showed that during storage,the dissolved oxygen content,ascorbic acid content,and sensory scores of the two products showed a decreasing trend over time,while the browning index,carbonyl content,and protein hydrolysis degree showed an increasing trend.There is no obvious pattern of changes in other indicators.Through correlation analysis,there is a significant correlation between browning index,carbonyl content,protein hydrolysis degree and sensory score,which can characterize the sensory quality changes of UHT milk during storage.(3)Establish a shelf life prediction model for UHT milk based on the comprehensive influence of multiple factors.Five types of UHT milk with different initial fat content and initial protein content were stored for 90 days at 23,30,and 37 ℃.The effects of different initial fat content,initial protein content,and storage temperature on the quality changes of UHT milk were analyzed.Based on the experimental results,the negative impact of storage temperature on the shelf life of UHT milk is significant.The initial fat content in UHT milk can slow down the quality deterioration reaction,while the initial protein content in UHT milk can accelerate the quality deterioration reaction.Combining the Pearson correlation coefficient,the browning index was determined as the key quality indicator for the shelf life prediction model of UHT milk,and its sensory threshold was determined to be 9.33.A multi factor comprehensive impact prediction model for UHT milk shelf life is constructed based on the Arrhenius equation combined with multiple linear regression methods.(4)Establishment of a shelf life prediction model for UHT milk/room temperature yogurt based on neural networks.The data set was established for the key quality indicators of UHT milk shelf life,and the prediction model of UHT milk shelf life based on the comprehensive influence of multiple factors of neural network was proposed and verified by means of multilayer perceptron(MLP)and back-propagation(BP)neural network.The deviation between the two models was small(MLP<10%,BP<1%).The performance of traditional dynamics,MLP neural network,and BP neural network was compared by combining linear fitting.The results showed that,The UHT milk multi factor shelf life model constructed by BP neural network has the highest fitting degree,with a correlation coefficient of 0.9957. |
