Improvement of blueberry freshness prediction based on machine learning and multi-source sensing in the cold chain logistics

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...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Food control Ročník 145; s. 109496
Hlavní autoři: Huang, Wentao, Wang, Xuepei, Zhang, Junchang, Xia, Jie, Zhang, Xiaoshuan
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Ltd 01.03.2023
Témata:
algorithms
automation
blueberries
Blueberry
carbon dioxide
cold chain
equations
food quality
freshness
Freshness prediction
fruits
HACCP
HACCP methods
Machine learning algorithm
prediction
temperature
Wireless sensor network (WSN)
Freshness prediction
Wireless sensor network (WSN)
HACCP methods
Blueberry
Machine learning algorithm
ISSN:0956-7135, 1873-7129
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí: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 (O2, CO2, and C2H4) 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. •A WSN-based environmental monitoring platform for blueberry cold chain was developed.•A model was developed to predict blueberry freshness at three different temperatures.•Non-destructive prediction of blueberry freshness using gas sensed information.•Gas information has a strong correlation with blueberry quality index.•The model proposed in this paper can be applied in cold chain logistics.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0956-7135
1873-7129
DOI:10.1016/j.foodcont.2022.109496