Modelo Predictivo de Humedad en Invernaderos Mediante Inferencia Difusa Aplicando Métodos de Optimización

Abstract

This document presents the results for prediction of internal relative humidity in greenhouses, allowing its behavior to be described in an easy, simple and interpretable way for people interested in agricultural activities, without requiring specific knowledge about handling data collection processing tools. Optimized fuzzy inference systems are defined using various methods and algorithms, which obtain the highest degree of interpretability and precision to facilitate monitoring and control, implemented using computational tools in greenhouse environments. The solution provides two models of fuzzy inference systems, depending on the data available in the greenhouse environment, optimized by means of a genetic and interior point algorithm applying a hybrid strategy, one of the models is based on historical data of relative humidity and uses systems ANFIS-type fuzzy inference systems, the other based on variables correlated with relative humidity from Mamdani-type fuzzy inference systems, both provide results with a high degree of precision and interpretability.