SISTEMA DE IRRIGAÇÃO INTELIGENTE BASEADO EM LÓGICA​ ​FUZZY INTEGRADO COM INTERNET DAS COISAS PARA A​ ​CULTURA DO TOMATE CEREJA​

Abstract

This work presents the development, implementation, and experimental validation of an Intelligent Irrigation System using Fuzzy Logic (SILF), integrated with the Internet of Things (IoT), designed to optimize water use in cherry tomato cultivation (Solanum lycopersicum var. cerasiforme). The proposed system addresses the limitations of traditional fixed-time irrigation methods, which often lead to under- or over-irrigation, nutrient leaching, and water waste.

The SILF architecture consists of: (i) a sensor network based on the ESP32 microcontroller for monitoring soil moisture, air temperature, and air humidity; (ii) a Python/Flask backend server hosting the fuzzy inference module; and (iii) a web-based interface developed with Angular for remote monitoring and system configuration. The core of the system is a Mamdani-type fuzzy controller, whose membership functions and inference rules were defined based on consultations with a fruit-growing specialist and on the crop’s physiological literature. Every hour, the system processes real-time environmental data, applies 27 fuzzy rules, and—using the centroid defuzzification method—determines the optimal activation time of the hydraulic pump and solenoid valve, ranging from 0 to 60 minutes.

Experimental validation was conducted in a protected environment using a randomized block design with a 5 × 3 factorial arrangement, comprising five substrate volumes (3, 6, 9, 12, and 15 L) and three irrigation methods: (i) traditional (30 continuous minutes once per day); (ii) fractionated (6 minutes every 2 hours, totaling 30 minutes per day); and (iii) intelligent irrigation using SILF. A total of 80 cherry tomato plants were evaluated, with drainage (excess water) as the primary response variable.

Analysis of variance (ANOVA) revealed highly significant effects (p < 0.0001) of both irrigation method and substrate volume on drainage. The interaction between these factors was not significant (p = 0.3494), indicating the robustness of SILF under different cultivation conditions. Tukey’s multiple comparison test (α = 0.05) demonstrated that SILF resulted in a statistically significant reduction in average drainage compared to traditional fixed-time irrigation methods, while no significant difference was observed between the two fixed-time strategies. Regarding substrate volume, pots with 15 L exhibited significantly lower drainage compared to those with 3 L and 9 L.

It is concluded that SILF is a viable and effective technological solution, capable of saving an average of 200 to 300 mL of water per plant per day, corresponding to a potential reduction of approximately 24 L per day for a bench containing 80 pots. The system combines low cost, scalability, remote control, and adaptive decision-making logic that mimics expert reasoning. This work validates the potential of integrating Fuzzy Logic and IoT technologies for precision agriculture, promoting a more sustainable use of water resources.