Articles
CROP MODEL REDUCTION AND SIMULATION IN REDUCED SPACE
Article number
406_5
Pages
63 – 72
Language
Abstract
TOMGRO is a rather complex, 69 state variable, crop growth model.
A linear reducing transformation, Principal Component Analysis, was used to find low-dimensional equivalents of the state vectors of TOMGRO. Dynamic Neural Networks (DNNs) were then trained to capture the dynamics of the reduced model.
Finally, the trained DNNs were used for simulations in reduced space.
The results of these simulations compared favorably with the original TOMGRO simulations.
A linear reducing transformation, Principal Component Analysis, was used to find low-dimensional equivalents of the state vectors of TOMGRO. Dynamic Neural Networks (DNNs) were then trained to capture the dynamics of the reduced model.
Finally, the trained DNNs were used for simulations in reduced space.
The results of these simulations compared favorably with the original TOMGRO simulations.
Publication
Authors
I. Seginer, I. Ioslovich
Keywords
Online Articles (48)