Yield prediction and growth mode characterization of greenhouse tomatoes with neural networks and fuzzy logic

E. Fitz-Rodríguez, Gene A Giacomelli

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

Despite the technological advances implemented in greenhouse crop production, greenhouse operation relies on human expertise to decide on the optimum values of each environmental control parameter. Most importantly, the selected values are determined by human observation of the crop responses. Greenhouse tomatoes often show a pattern of cycling between reproductive and vegetative growth modes. The growth mode is a practical visual characterization of the source-sink relationships of the plants resulting from the greenhouse environment (aerial and root zone). Experienced reenhouse tomato growers assess the growth mode based on morphological observations, including quantitative (length, diameter, elongation rates) and qualitative (shape and color) features of the plant head, stems, flowers, trusses, and leaves. Data from greenhouse environments and crop records from an experimental production in Tucson, Arizona, and from a large-scale commercial operation in Marfa, Texas, were used for modeling the growth mode of tomato plants with fuzzy logic. Data from the commercial operation were used to model weekly fluctuations of harvest rate, fruit size, and fruit developing time with dynamic neural networks (NN). The NN models accurately predicted weekly and seasonal fluctuations of the fruit-related parameters, having coefficients of determination(R2) of 0.92, 0.76, and 0.88, respectively, for harvest rate, fruit fresh weight, and fruit developing time, when compared with a dataset used for independent validation. The fuzzy modeling of growth mode allowed discrimination of the reproductive and balanced growth modes in the experimental system, and modeling of the seasonal growth mode variation in the commercial application. Both modeling results might be applicable to commercial operations for making decisions on greenhouse climate control and overall crop management practices.

Original languageEnglish (US)
Pages (from-to)2115-2128
Number of pages14
JournalTransactions of the ASABE
Volume52
Issue number6
StatePublished - 2009

Fingerprint

Fuzzy Logic
fuzzy logic
Greenhouses
fuzzy mathematics
Lycopersicon esculentum
neural networks
Fuzzy logic
Fruits
tomatoes
Neural networks
greenhouses
Fruit
prediction
fruit
Growth
Crops
fruits
crop
modeling
Trusses

Keywords

  • Dynamic neural network
  • Fuzzy logic
  • Greenhouse tomato
  • Growth mode
  • Modeling
  • Pattern recognition
  • Reproductive
  • Vegetative

ASJC Scopus subject areas

  • Agronomy and Crop Science
  • Biomedical Engineering
  • Food Science
  • Forestry
  • Soil Science

Cite this

Yield prediction and growth mode characterization of greenhouse tomatoes with neural networks and fuzzy logic. / Fitz-Rodríguez, E.; Giacomelli, Gene A.

In: Transactions of the ASABE, Vol. 52, No. 6, 2009, p. 2115-2128.

Research output: Contribution to journalArticle

@article{3835bda444b94ddc95a4ed9933a03235,
title = "Yield prediction and growth mode characterization of greenhouse tomatoes with neural networks and fuzzy logic",
abstract = "Despite the technological advances implemented in greenhouse crop production, greenhouse operation relies on human expertise to decide on the optimum values of each environmental control parameter. Most importantly, the selected values are determined by human observation of the crop responses. Greenhouse tomatoes often show a pattern of cycling between reproductive and vegetative growth modes. The growth mode is a practical visual characterization of the source-sink relationships of the plants resulting from the greenhouse environment (aerial and root zone). Experienced reenhouse tomato growers assess the growth mode based on morphological observations, including quantitative (length, diameter, elongation rates) and qualitative (shape and color) features of the plant head, stems, flowers, trusses, and leaves. Data from greenhouse environments and crop records from an experimental production in Tucson, Arizona, and from a large-scale commercial operation in Marfa, Texas, were used for modeling the growth mode of tomato plants with fuzzy logic. Data from the commercial operation were used to model weekly fluctuations of harvest rate, fruit size, and fruit developing time with dynamic neural networks (NN). The NN models accurately predicted weekly and seasonal fluctuations of the fruit-related parameters, having coefficients of determination(R2) of 0.92, 0.76, and 0.88, respectively, for harvest rate, fruit fresh weight, and fruit developing time, when compared with a dataset used for independent validation. The fuzzy modeling of growth mode allowed discrimination of the reproductive and balanced growth modes in the experimental system, and modeling of the seasonal growth mode variation in the commercial application. Both modeling results might be applicable to commercial operations for making decisions on greenhouse climate control and overall crop management practices.",
keywords = "Dynamic neural network, Fuzzy logic, Greenhouse tomato, Growth mode, Modeling, Pattern recognition, Reproductive, Vegetative",
author = "E. Fitz-Rodr{\'i}guez and Giacomelli, {Gene A}",
year = "2009",
language = "English (US)",
volume = "52",
pages = "2115--2128",
journal = "Transactions of the ASABE",
issn = "2151-0032",
publisher = "American Society of Agricultural and Biological Engineers",
number = "6",

}

TY - JOUR

T1 - Yield prediction and growth mode characterization of greenhouse tomatoes with neural networks and fuzzy logic

AU - Fitz-Rodríguez, E.

AU - Giacomelli, Gene A

PY - 2009

Y1 - 2009

N2 - Despite the technological advances implemented in greenhouse crop production, greenhouse operation relies on human expertise to decide on the optimum values of each environmental control parameter. Most importantly, the selected values are determined by human observation of the crop responses. Greenhouse tomatoes often show a pattern of cycling between reproductive and vegetative growth modes. The growth mode is a practical visual characterization of the source-sink relationships of the plants resulting from the greenhouse environment (aerial and root zone). Experienced reenhouse tomato growers assess the growth mode based on morphological observations, including quantitative (length, diameter, elongation rates) and qualitative (shape and color) features of the plant head, stems, flowers, trusses, and leaves. Data from greenhouse environments and crop records from an experimental production in Tucson, Arizona, and from a large-scale commercial operation in Marfa, Texas, were used for modeling the growth mode of tomato plants with fuzzy logic. Data from the commercial operation were used to model weekly fluctuations of harvest rate, fruit size, and fruit developing time with dynamic neural networks (NN). The NN models accurately predicted weekly and seasonal fluctuations of the fruit-related parameters, having coefficients of determination(R2) of 0.92, 0.76, and 0.88, respectively, for harvest rate, fruit fresh weight, and fruit developing time, when compared with a dataset used for independent validation. The fuzzy modeling of growth mode allowed discrimination of the reproductive and balanced growth modes in the experimental system, and modeling of the seasonal growth mode variation in the commercial application. Both modeling results might be applicable to commercial operations for making decisions on greenhouse climate control and overall crop management practices.

AB - Despite the technological advances implemented in greenhouse crop production, greenhouse operation relies on human expertise to decide on the optimum values of each environmental control parameter. Most importantly, the selected values are determined by human observation of the crop responses. Greenhouse tomatoes often show a pattern of cycling between reproductive and vegetative growth modes. The growth mode is a practical visual characterization of the source-sink relationships of the plants resulting from the greenhouse environment (aerial and root zone). Experienced reenhouse tomato growers assess the growth mode based on morphological observations, including quantitative (length, diameter, elongation rates) and qualitative (shape and color) features of the plant head, stems, flowers, trusses, and leaves. Data from greenhouse environments and crop records from an experimental production in Tucson, Arizona, and from a large-scale commercial operation in Marfa, Texas, were used for modeling the growth mode of tomato plants with fuzzy logic. Data from the commercial operation were used to model weekly fluctuations of harvest rate, fruit size, and fruit developing time with dynamic neural networks (NN). The NN models accurately predicted weekly and seasonal fluctuations of the fruit-related parameters, having coefficients of determination(R2) of 0.92, 0.76, and 0.88, respectively, for harvest rate, fruit fresh weight, and fruit developing time, when compared with a dataset used for independent validation. The fuzzy modeling of growth mode allowed discrimination of the reproductive and balanced growth modes in the experimental system, and modeling of the seasonal growth mode variation in the commercial application. Both modeling results might be applicable to commercial operations for making decisions on greenhouse climate control and overall crop management practices.

KW - Dynamic neural network

KW - Fuzzy logic

KW - Greenhouse tomato

KW - Growth mode

KW - Modeling

KW - Pattern recognition

KW - Reproductive

KW - Vegetative

UR - http://www.scopus.com/inward/record.url?scp=76249099775&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=76249099775&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:76249099775

VL - 52

SP - 2115

EP - 2128

JO - Transactions of the ASABE

JF - Transactions of the ASABE

SN - 2151-0032

IS - 6

ER -