Rancang Bangun Sistem Monitoring Kematangan Pisang Raja Nangka Fase Pascapanen Dengan Analisis Tren Gas Berbasis IoT

Haffian Naufal Afrizal, Ir. Muhammad Arrofiq, S.T., M.T. Ph.D.

2026 | Tugas Akhir | D4 TEKNOLOGI JARINGAN

Pisang raja nangka memiliki karakteristik stay-green sehingga kematangannya sulit diidentifikasi berdasarkan warna kulit. Penelitian ini merancang sistem pemantauan kematangan pisang pascapanen noninvasif berbasis Internet of Things (IoT) menggunakan analisis tren gas etanol. Perangkat keras meliputi mikrokontroler ESP32, sensor gas MQ3, dan DHT22 di dalam kontainer tertutup. Klasifikasi dilakukan langsung pada mikrokontroler menggunakan algoritma Consecutive Monotone Trend Detection yang mengevaluasi nilai Moving Average 5 data (MA5) dan laju perubahannya (Slope5). Pemantauan berlangsung real-time menggunakan bot Telegram yang terhubung ke Virtual Private Server (VPS) dan MariaDB melalui protokol MQTT. Hasil pengujian tiga Batch membuktikan sistem berhasil mengklasifikasikan empat fase kematangan (Warm_up, Mentah, Mulai_Matang, dan Matang). Ambang batas empiris yang diperoleh yaitu MA5 di bawah 900 ADC untuk Mentah, MA5 minimal 900 ADC atau Slope5 minimal 20 untuk Mulai_Matang, dan MA5 minimal 2200 ADC atau Slope5 minimal 30 untuk Matang. Ambang batas ini diuji pada tiga Batch dengan 603 data pada Batch 1, 477 data pada Batch 2, serta 564 dan 513 data pada dua kontainer Batch 3, dan tetap konsisten mengklasifikasikan tingkat kematangan tanpa penyetelan ulang.

Raja Nangka bananas have a stay-green characteristic, making their ripeness difficult to identify based on skin color. This study developed a non-invasive Internet of Things (IoT) based monitoring system for post-harvest banana ripeness using ethanol gas trend analysis. The hardware consists of an ESP32 microcontroller, an MQ3 gas sensor, and a DHT22 sensor housed inside a closed container. Classification is performed on the microcontroller using the Consecutive Monotone Trend Detection algorithm, which evaluates the 5-data-point Moving Average (MA5) and its rate of change (Slope5). Monitoring is conducted in real time using a Telegram bot connected to a Virtual Private Server (VPS) and MariaDB through the MQTT protocol. Test results from three Batches demonstrate that the system successfully classified four ripeness phases (Warm_up, Unripe, Starting_to_Ripen, and Ripe. The empirical thresholds obtained are MA5 below 900 ADC for “Raw,” MA5 of at least 900 ADC or Slope5 of at least 20 for “Starting to Ripen,” and MA5 of at least 2,200 ADC or Slope5 of at least 30 for “Ripe.” These thresholds were tested on three Batches with 603 data points in Batch 1, 477 data points in Batch 2, and 564 and 513 data points in the two containers of Batch 3, and consistently classified the ripeness levels without the need for readjustment.

Kata Kunci : IoT, sensor gas, MQ3, pisang raja nangka, deteksi kematangan, moving average, rule-based, pascapanen, ESP32, MQTT