Sensor, IoT-based post-harvest shelf life determination of tomato (Lycopersicon esculentum) through machine learning predictive analysis for intelligent transport

Aim: The current research explores the potential of machine learning predictive models in optimizing the storage conditions of tomatoes. This is achieved through Internet of Things (IoT) technology, sensors, cameras, and microprocessors integrated into refrigerators along the supply chain. Methodolo...

Full description

Saved in:
Bibliographic Details
Published in:Journal of environmental biology Vol. 45; no. 4; pp. 455 - 464
Main Authors: Shankaraswamy, J., Radhika, T.S.L.
Format: Journal Article
Language:English
Published: Lucknow Triveni Enterprises 01.07.2024
Subjects:
Artificial intelligence
Carbon
Commodities
Data collection
Effectiveness
Food
Food quality
Fruits
Harvest
Horticulture
Humidity
Internet of Things
Learning algorithms
Lycopersicon esculentum
Machine learning
Microcontrollers
Microprocessors
Optimization
Oxygen
Physiology
Polynomials
Prediction models
Sensors
Shelf life
Software
Software development tools
Storage conditions
Supply chains
Temperature
Tomatoes
Vegetables
India
ISSN:0254-8704, 2394-0379
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Aim: The current research explores the potential of machine learning predictive models in optimizing the storage conditions of tomatoes. This is achieved through Internet of Things (IoT) technology, sensors, cameras, and microprocessors integrated into refrigerators along the supply chain. Methodology: Controlling temperature and humidity inside the refrigerated container was accomplished by implementing the Arduino microcontroller and supplementary hardware components, including the ESP32 module relay, an advancement over the ESP8266 microcontroller. The Arduino Integrated Development Environment (IDE) was used as software platform for this experimentation. Various parameters, including humidity, oxygen, carbon-di-oxide, and shelf life, were recorded at different temperatures and on different days. Subsequently, the collected data was analyzed employing machine-learning models to determine the most effective prediction model for these variables. Results: From the results it has been revealed that apolynomial of degree 4 is the best-fit regressor model for the data on humidity. Polynomials of degrees 2, 2, and 3 are the best models for the target variables oxygen, carbon-di-oxide, and shelf life. Interpretation: During analysis, This result suggests that different polynomial degrees are optimal for modeling different variables in the dataset. Polynomials of degrees 2, 2, and 3 are the best ML models for the target variables oxygen, carbon-di-oxide, and shelf life, respectively,to enhance the effectiveness of our predictive models. Key words: Io T sensors, ML models, Quantile loss, Supply chain, Tomato
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0254-8704
2394-0379
DOI:10.22438/jeb/45/4/MRN-5339