A DYNAMIC SIMULATION MODEL FOR SEEDLING GROWTH

Abstract

This study presents a dynamic growth model applicable to automated seedling cultivation. Experiments on the influence of environmental conditions on cabbage seedling quality during three growth stages were conducted in a phytotron, and a growth database was established. An error back propagation neural network was used to analyze experimental data and develop strategies for a dynamic growth model to simulate the relationship between environmental factors (temperature, water supply and daily radiation) and cabbage seedling quality (cumulative dry matter of seedlings). A feedback algorithm and dynamic strategies were integrated into the neural network to reflect the strong importance of daily historical memory in seedling growth. The dynamic model was thus successfully developed with a coefficient of determination of 0.996 and error of 1.68%, and was verified using the data from nurseries. The dynamic model performed excellently in determining seedling growth, achieving superior results to static models. The error in predicting the cumulative dry matter resulting from seedling growth was reduced by about 80% (from 18.2% to 3.75% prediction error) when the dynamic growth model was used in place of the static model. This model not only gave a clear view of production management toward seedling growth, but also provided a basis for better environmental and quality control strategies.