Rancang Bangun Alat Penyimpan dan Pendeteksi Kematangan Pisang Ambon (Musa paradisiaca var. sapientum) Menggunakan Sensor TCS34725 Berbasis Machine learning

Abstract

Banana is one of the most widely produced fruits in Indonesia; however, it is highly perishable after harvest. One environmental factor that needs to be considered to maintain post-harvest fruit quality is storage temperature. This study aims to design a prototype banana storage system with automatic temperature control and banana ripeness detection using a TCS34725 sensor implemented with machine learning. The system is equipped with a DHT22 sensor to monitor temperature and humidity, an MQ3 sensor to detect alcohol content, and a Peltier element to control the storage room temperature. The ripeness level classification uses a machine learning model trained with the SVM method with 94% accuracy. The test results show that the system is capable of maintaining the storage temperature in the range of 20°C to 25°C. The test
was conducted using 5 bananas placed inside the system and 5 bananas outside the system. The bananas stored in the system ripened on the 10th day, while the bananas stored outside the system ripened on the 8th day. The bananas in the system had a shelf life of up to 23 days before rotting, while the bananas stored outside the system rotted on the 19th day. The bananas stored in the system experienced a 31% weight loss, while the bananas stored outside the system experienced a 38% weight loss. The system is capable of displaying real-time banana conditions through an LCD interface.