Mixed particle swarm optimization algorithm-based approach to optimize spatial distribution of virtual maize

Abstract

Increasing crop yields is one potential way to solve the problem of food shortage and limited field resources. Optimization of crop spatial distribution is an efficient method to increase yield. However, extensive field experiments are costly and time consuming, which renders conventional optimization studies difficult if not prohibitive. In this study, a new optimization strategy was proposed, integrating a Functional-Structural Plant Model (FSPM) of maize with a workflow based on a Mixed Particle Swarm Optimization (MPSO) algorithm. The maize FSPM applies a dynamic interactive feedback method between plant morphology and plant growth dynamics on the one hand, and simultaneously between plant spatial distribution and plant growth, and furthermore can respond to different environmental factors. MPSO is a new mixed particle swarm optimization algorithm with multistage disturbance, which enhances the ability of traversing the solution space and jumping out of local extrema. This study performs a plant spatial distribution optimization of monocropping maize to maximize light interception. Simulation results indicate that the optimized row spacing can increase maize light interception comparing to the random row spacing within the spacing range (with maximum of 36.44% increment), and that the optimization results satisfy the optimal interval provided by previous field experiments.