220314 Mon Thesis abstract

Studies of evolutionary algorithms for the reduced Tomgro model calibration for modelling tomato yields

 

Gong et al., 2021

Smart Agricultrual Technology

 

Thanks to technology development, one can control greenhouse environment to make optimal condition for maximum crop yield. While data driven models acquire parameters from large,historical data, biophysical models describe the data mathematically, which makes it practical to apply. There have been many biophysical models attempted to explain relationship between crop yield and other environmental factors. For tomatoes, TOMSIM and TOMGRO were developed and validated so far, and tomgro was fairly accurate. Despite their variations, their massive amount of variables make it difficult to apply the model. Therefore, simplified Tomgro model was suggested in 1999, shoed successful simulation results in various greenhouses. Even though many evolutionary algorithms for model calibration exist, there was rarely any research applying the reduced tomgro model. Therefore, evolutionary algorithms called genetic algorithm, partical swarm optimization(PSO) algorithm and differential evolutionary algorithms were used for tomato dry matter calibration, and three different error metrics were used to compare them. PSO turned out to be the most precise calibration algorithm.