Fuzzy Logic and Artificial Neural Network Perceptron Multi-Layer and Radial Basis in Estimating Marandu Grass Yield in Integrated Systems

Abstract

ABSTRACT The integrated crop-livestock-forest (ICLF) system integrates different components of animal husbandry. The implementation of this system aims at sustainability, seeking to exploit the area as much as possible, in addition to reducing the impact on the physical, chemical, and biological properties of the soil. With technological advances and numerous variables, fuzzy logic, and artificial neural networks (ANNs) have been used for data classification and estimation. This study aims to estimate the Marandu grass yield in integrated systems using the input, volume of rainfall, and experimental period. A performance of approximately 0.077 was observed for the mean square error (MSE), and the radial basis in estimation (RBR) network had an error of 0.255%, which is much lower than that of the multi-layer perceptron (MLP) network and methodology based on fuzzy logic, with errors of 2.713 and 10.840%, respectively, between the obtained and expected output. This indicates that the quality of the grass was better with one or three eucalyptus lines in the ICLF system and demonstrates the application efficiency of the model with a tool for forecasting the Marandu grass yield in the studied soil and climate conditions.