Implementasi Early Warning System (EWS) Menggunakan Prediksi Metode Sarima pada Modul Environmental Monitoring System (EMS) Berbasis IoT

Abstract

Penelitian ini mengembangkan Environmental Monitoring System (EMS) berbasis Internet of Things (IoT) yang dilengkapi fitur Early Warning System (EWS) menggunakan metode Seasonal Autoregressive Integrated Moving Average (SARIMA) untuk meningkatkan keandalan pemantauan ruang server. Sistem monitoring konvensional yang bersifat reaktif dinilai kurang efektif dalam mengurangi risiko downtime karena peringatan baru diberikan setelah kondisi kritis terjadi. Sistem yang dikembangkan memantau suhu, kelembapan, potensi api, dan response time server secara real-time menggunakan mikrokontroler ESP8266 dan sensor pendukung, kemudian menyimpan data historis ke dalam basis data. Model SARIMA dilatih menggunakan data historis untuk menghasilkan prediksi kondisi hingga dua jam ke depan sebagai dasar pemberian peringatan dini. Hasil implementasi menunjukkan bahwa sistem mampu mengintegrasikan proses pemantauan dan prediksi secara otomatis sehingga dapat memberikan peringatan sebelum parameter mencapai ambang batas kritis

This study develops an Internet of Things (IoT)-based Environmental Monitoring System (EMS) equipped with an Early Warning System (EWS) using the Seasonal Autoregressive Integrated Moving Average (SARIMA) method to improve the reliability of server room monitoring. Conventional monitoring systems are generally reactive, providing alerts only after critical conditions have occurred, making them less effective in reducing the risk of server downtime. The proposed system monitors temperature, humidity, fire potential, and server response time in real time using an ESP8266 microcontroller and supporting sensors, while storing historical data in a database. The SARIMA model is trained using historical data to predict environmental conditions up to two hours in advance, providing the basis for proactive early warnings. The implementation results demonstrate that the system successfully integrates real-time monitoring with predictive analytics, enabling early warnings before monitored parameters reach critical thresholds