Modelo matemático para la predicción del crecimiento del Brassica oleracea var, itálica (brócoli) en el invernadero 1, de la Finca Freire del cantón Latacunga basado en redes neuronales.

Abstract

The objective of this paper was to design a mathematical model to predict the growth of Legacy broccoli in the Latacunga canton using neural networks, a support tool known as simulation that allows improving the production process, taking into consideration 3 different scenarios: in the first scenario, plantation with normal standards, while in the second and third scenarios the chemical variations are analyzed to observe if the product adapts to these circumstances. The mathematical model is based on the three-layer perceptron neural network. The first layer has 12 parameters, the second layer has 10 neurons and the output layer has two parameters that help identify the prediction of the growth of Legacy broccoli. Using the Matlab program, the growth prediction is obtained, resulting in the verification of the hypothesis of the fifth scenario (simulated scenario), the element calcium was altered to 501 ppm, obtaining a stem height of 53.97 cm and a foil diameter of 17.73. It was, at this point, proven that increasing calcium by 5ppm, broccoli develops better. Finally, the mathematical model is validated by Mean Absolute Percentage Error, using real data and simulated data, resulting in an error of 2.936%.