Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/165093| Title: | Pengembangan Sistem Deteksi Asap dan Api Berbasis IoT Menggunakan Analisis Random Forest Regressor |
| Other Titles: | Development of IoT-Based Smoke and Fire Detection System Using Random Forest Regressor Analysis |
| Authors: | Wahyudi, Setyanto Tri Herti |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Kebakaran dan keberadaan asap di dalam ruangan tertutup dapat menyebabkan kerusakan material yang signifikan serta membahayakan kesehatan manusia. Untuk mengantisipasi hal tersebut, penelitian ini bertujuan merancang sistem deteksi dini berbasis Internet of Things (IoT) yang mampu mendeteksi keberadaan asap dan api secara akurat
dan responsif. Sistem ini menggunakan sensor MQ-2 untuk mendeteksi asap dan sensor Flame untuk mendeteksi api, yang dikendalikan oleh mikrokontroler ESP32 serta diintegrasikan dengan bot Telegram guna mengirimkan notifikasi secara real-time saat potensi bahaya terdeteksi. Evaluasi performa sensor MQ-2 dilakukan dengan membandingkan data hasil pembacaan terhadap alat kalibrator menggunakan algoritma Random Forest Regressor. Hasil pengujian menunjukkan bahwa sensor MQ-2 memiliki tingkat akurasi sebesar 98% dengan nilai galat (MSE) sebesar 0,192 dan nilai koefisien determinasi (R²) sebesar 0,999. Sementara itu, sensor Flame menunjukkan performa optimal pada jarak 70 cm dan sudut 90° dengan sumber api dari korek api, namun
sensitivitasnya menurun pada sudut ekstrem 30° dan 120°, serta kehilangan kemampuan deteksi setelah melewati jarak 80 cm. Pengujian menggunakan sumber api yang lebih besar
berupa kertas A5 menunjukkan peningkatan jangkauan deteksi hingga 200 cm, dengan efektivitas tetap terjaga pada sudut 60° hingga 120. Fire and the presence of smoke in a closed room can cause significant material damage and endanger human health. To anticipate this, this research aims to design an Internet of Things (IoT)-based early detection system that is able to detect the presence of smoke and fire accurately and responsively. The system uses an MQ-2 sensor to detect smoke and a Flame sensor to detect fire, which is controlled by an ESP32 microcontroller and integrated with a Telegram bot to send real-time notifications when potential danger is detected. The performance evaluation of the MQ-2 sensor is carried out by comparing the reading data against the calibrator tool using the Random Forest Regressor algorithm. The test results show that the MQ-2 sensor has an accuracy rate of 98% with an error value (MSE) of 0.192 and a coefficient of determination (R²) of 0.999. Meanwhile, the Flame sensor showed optimal performance at a distance of 70 cm and an angle of 90° with a fire source from a lighter, but its sensitivity decreased at extreme angles of 30° and 120°, and lost detection capability after passing a distance of 80 cm. Tests using a larger flame source of A5 paper showed an increased detection range of up to 200 cm, with effectiveness maintained at angles of 60° to 120°. |
| URI: | http://repository.ipb.ac.id/handle/123456789/165093 |
| Appears in Collections: | UT - Computer Engineering Tehcnology |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_J0304211175_6bf17c8b9c344e339a60209e54245dfd.pdf | Cover | 481.77 kB | Adobe PDF | View/Open |
| fulltext_J0304211175_b1369faf82024923917b2a7364ee2d7e.pdf Restricted Access | Fulltext | 1.51 MB | Adobe PDF | View/Open |
| lampiran_J0304211175_efec5d548c224109bdafe458cb952208.pdf Restricted Access | Lampiran | 412.11 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.