PARAMETER ESTIMATION FOR CROP GROWTH MODEL USING EVOLUTIONARY AND BIO-INSPIRED ALGORITHMS

The calibration of dynamic models for crop growth and development with parameter ranges generates imprecise estimations and erroneous predictions of the model when estimation is used by least squares or maximum likelihood. The present study shows the use of global methods of optimization for resolving this problem. A comparison is presented of the performance of an evolutionary algorithm (Differential Evolution, DE) and two that are bio-inspired: Cuckoo Search (CS) and Modified Cuckoo Search (MCS). The test problem was to estimate the 25 parameters of the model for potential crop growth SUCROS (a Simple and Universal Crop growth Simulator).The data used was obtained from an experiment of growth of a husk tomato crop (Physalis ixocalpa Brot. Ex Horm.) carried out at Chapingo, Mexico. The objective was to determine which algorithm generates values for the parameters of the model that make it possible to obtain the most precise predictions. An ANOVA was performed to statistically evaluate the efficiency and effectiveness of the proposed algorithms. Results showed a better performance of the standard DE algorithm (DE/rand/1/bin) in terms of efficiency and effectiveness to converge to an optimum solution. The bio-inspired algorithms showed good performance; therefore, they are reliable and can be applied in the estimation of parameters of crop models.