Improvement of blueberry freshness prediction based on machine learning and multi-source sensing in the cold chain logistics
Wentao Huang,Xuepei Wang,Junchang Zhang,Jie Xia,Xiaoshuan Zhang
DOI: https://doi.org/10.1016/j.foodcont.2022.109496
IF: 6
2022-11-17
Food Control
Abstract:Traditional fruit freshness prediction and modeling heavily rely on various physicochemical indicators (such as water loss rate, pH, and VC content), which is facing predicaments of time-consuming, laborious, destructive, and low prediction accuracy. To this end, this paper proposes a new method for fruit freshness prediction based on multi-sensing technology and machine learning algorithm, thereby improving the automation, intelligentialize, and high accuracy of fruit freshness prediction. The critical control points of blueberry cold chain logistics were analyzed firstly based on the HACCP method, identifying the key gas parameters (O 2 , CO 2 , and C 2 H 4 ) and interaction mechanisms of gas and blueberry freshness. Then the blueberry cold chain microenvironment monitoring platform (BCCMMP) was developed for critical gas content monitoring at different temperatures (0 °C, 5 °C, and 22 °C). It was demonstrated that gas information can replace quality information to characterize blueberry freshness, and further emerging machine learning (ML) algorithms (BP, RBF, SVM, and ELM) were constructed for blueberry freshness prediction using critical gas information, and the results showed prediction accuracies of 90.87% (BP), 92.24% (RBF), 94.01% (SVM), and 91.31% (ELM). By contrast, the 85.10% prediction accuracy was achieved by the traditional Arrhenius equation method based on temperature and quality parameters. Through the automatic non-destructive acquisition of sensing data and emerging machine learning algorithms, this paper provides a new approach to improving the freshness prediction accuracy and food quality management level during fruit cold chain logistics.
food science & technology