Pembuatan Sistem Deteksi Kebakaran Berbasis IoT Menggunakan Algoritma Decision Tree di RIM Telkom Corporate University

Abstract

Kebakaran merupakan bencana yang dapat menyebabkan kerugian material maupun korban jiwa sehingga diperlukan sistem deteksi dini yang mampu memberikan informasi secara cepat dan akurat. Penelitian ini bertujuan merancang dan mengimplementasikan sistem deteksi kebakaran berbasis Internet of Things (IoT) menggunakan algoritma Decision Tree C4.5. Sistem dikembangkan menggunakan mikrokontroler ESP32-S3 yang terintegrasi dengan sensor suhu, karbon monoksida (CO), dan api. Data sensor ditampilkan pada website monitoring secara real-time. Dataset terdiri atas 2.000 data yang dibagi dengan rasio 80:20 untuk pelatihan dan pengujian model. Hasil kalibrasi menunjukkan rata-rata error sensor MQ-7 sebesar 1,12% dan DHT22 sebesar 1,13%. Model Decision Tree C4.5 menghasilkan nilai accuracy 95,75%, precision 95,79%, recall 95,63%, dan F1-score 95,70%. Hasil penelitian menunjukkan bahwa sistem mampu mendeteksi dan mengklasifikasikan kondisi ruangan menjadi Aman, Waspada, dan Bahaya.

Fire can cause significant material losses and casualties, making an early detection system essential for providing timely and accurate information. This study aims to design and implement an Internet of Things (IoT)-based fire detection system using the Decision Tree C4.5 algorithm. The system was developed using an ESP32-S3 microcontroller integrated with temperature, carbon monoxide (CO), and flame sensors. Sensor data were displayed on a real-time web-based monitoring platform. The dataset consisted of 2,000 records divided with an 80:20 ratio for model training and testing. Calibration results showed average errors of 1.12% for the MQ-7 sensor and 1.13% for the DHT22 sensor. The Decision Tree C4.5 model achieved an accuracy of 95.75%, precision of 95.79%, recall of 95.63%, and F1-score of 95.70%. The results indicate that the system can effectively detect and classify room conditions into Safe, Alert, and Danger.