PERANCANGAN SISTEM IRIGASI BERBASIS INTERNET OF THINGS (IOT) MENGGUNAKAN ALGORITMA RECURRENT NEURAL NETWORK (RNN) UNTUK TANAMAN CABAI RAWIT (CAPSICUM FRUTESCENS L)

Irrigation systems are critical components in improving agricultural productivity, especially in the cultivation of chili peppers that require precise water management. This research aims to design an automatic Internet of Things (IoT)-based irrigation system by integrating weather prediction using Recurrent Neural Network (RNN) algorithm.
The research methodology includes developing an IoT system using hardware such as Wemos D1 R2, DHT11 sensors for monitoring air temperature and humidity, soil moisture sensors for measuring soil moisture, and utilizing weather data from Visual Crossing API. The RNN model was developed using historical weather data from Sukabumi for the period 2022-2024, focusing on four main parameters: temperature, humidity, precipitation, and wind speed.
The RNN model architecture was designed with a 64-unit SimpleRNN layer, a 32-unit Dense layer, and ReLU and softmax activation functions for weather condition classification. Research results show the model can predict weather conditions with 92% accuracy, validating the effectiveness of the machine learning approach in irrigation systems.
The automatic irrigation system successfully integrates weather prediction to optimize watering, resulting in a significant reduction in water wastage and improved irrigation efficiency. The system can perform watering at 08:15 and 15:15 based on weather prediction. For example, if the weather prediction shows clear conditions, the Wemos will set watering for 6 seconds, which is equivalent to 400 ml per day for one plant. The implementation of the MQTT protocol ensures real-time communication between the hardware and the web platform.
The primary contribution of this research is presenting an innovative technological solution supporting precision agriculture, particularly chili pepper cultivation, by leveraging artificial intelligence and Internet of Things for sustainable water resource management.
Keywords: Internet of Things, Recurrent Neural Network, Irrigation System, Chili Plants, Weather Prediction