Monitoring Suhu Chiller berbasis Internet of Things Menggunakan Thermal Camera

Arlan, Nurul Akbar (2023) Monitoring Suhu Chiller berbasis Internet of Things Menggunakan Thermal Camera. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07111940000075-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (9MB) | Request a copy

Abstract

Cold Chain dideskripsikan sebagai alat-alat dan proses yang melakukan suatu cara untuk memastikan produk terjaga suhunya hingga sampai ke konsumen. Produk ini bersifat mudah busuk, rentan, dan memiliki jangka waktu yang pendek. Cold Chain Logistic merupakan gabungan dari empat sistem yang bekerja sama yaitu precooling, gudang pendingin, transportasi dengan pendinginan, dan pemasaran. Kelemahan sistem pemantauan yang digunakan sekarang ialah menggunakan sensor yang tidak merepresentasikan suhu satu ruangan sehingga diperlukan Kamera Termal untuk mendeteksi suhu dari produk-produk tersebut karena dapat mendeteksi suhu di banyak titik sehingga lebih merepresentasikan satu ruangan chiller. Metode penelitian yang digunakan pada penelitian ini adalah pembuatan alat untuk memantau dan mendeteksi objek di dalam chiller dengan menggunakan Computer Vision untuk menampilkan gambar termal yang ditangkap oleh kamera termal, lalu Object Detection dengan model YOLOv5 untuk mendeteksi objek yang akan dipantau, dan Internet of Things untuk perpindahan datanya. Alat ini dibuat menggunakan sensor MLX90640 yang terhubung dengan ESP32 dan dikirim ke Personal Computer untuk diproses oleh YOLOv5 sehingga dapat mendeteksi objek dan pemantauan suhu dengan sistem peringatan dini. Setelah dilakukan penelitian didapat sensor MLX90640 dapat mengukur suhu ruangan Chiller dengan persamaan polinomial orde dua yang didapat dari proses kalibrasi, hasil pengukuran oleh sensor MLX90640 diproses dengan Computer Vision dan divisualisasikan menjadi gambar termal. Gambar termal diproses dengan model YOLOv5x yang telah dilatih, Hasil pelatihan menghasilkan model YOLOv5 dengan menghasilkan nilai mAP terbaik sebesar 0.70017, Object Loss terkecil sebesar 0.012879, dan Box Loss terkecil sebesar 0.016463 dengan Precision sebesar 0.97603 dan Recall sebesar 0.96536. Pengukuran suhu didapat dari Bounding Box hasil deteksi objek yang dihitung rata-rata suhu di dalamnya dengan ukuran 10% dari luas kotak untuk objek botol dan 30% dari luas kotak untuk objek vial dan sistem peringatan dini berupa alarm yang aktif ketika suhu objek atau ruangan melebihi batas yang telah diatur. Data-data suhu dikirimkan ke server Thingspeak untuk ditampilkan pada Dashboard Thingspeak.
================================================================================================================================
Cold Chain is described as tools and processes that carry out a method to ensure that the product is maintained at the temperature until it reaches the consumer. This product is perishable, vulnerable, and has a short shelf life. Cold Chain Logistics is a combination of four systems that work together, namely precooling, cold storage, refrigerated transportation, and marketing. The weakness of the current monitoring system is that it uses a sensor that does not represent the temperature of one room, so a Thermal Camera is needed to detect the temperature of these products because it can detect temperatures at many points so that it is more representative of one chiller room. The research method used in this study is the creation of a tool to monitor and detect objects in the chiller using Computer Vision to display thermal images captured by thermal cameras, then Object Detection with the YOLOv5 model to detect objects to be monitored, and the Internet of Things to data transfer. This tool is made using the MLX90640 sensor connected to ESP32 and sent to the Personal Computer to be processed by YOLOv5 so that it can detect objects and monitor the temperature with an early warning system. After doing research, it was found that the MLX90640 sensor can measure room temperature Chiller with the second order polynomial equation from calibration process, the measurement results by the MLX90640 sensor are processed with Computer Vision and visualized into thermal images. Thermal images were processed with the YOLOv5x model that had been trained. The results of the training produced the YOLOv5 model with the best mAP value of 0.70017, the smallest Object Loss of 0.012879, and the smallest Box Loss of 0.016463 with a Precision of 0.97603 and Recall of 0.96536. Temperature measurements are obtained from the Bounding Box resulting from object detection, which calculates the average temperature inside with a size of 10% of the box area for bottle objects and 30% of the box area for vial objects and an early warning system in the form of an active alarm when the object or room temperature exceeds a set limit. Temperature data is sent to the Thingspeak server to be displayed on the Thingspeak Dashboard.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Chiller, Cold Chain Logistric, Internet of Things, Thermal Camera, Object Detection
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5103.2 Wireless communication systems. Two way wireless communication T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7871.674 Detectors. Sensors T Technology > TR Photography
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Nurul Akbar Arlan
Date Deposited:	24 Jul 2023 01:45
Last Modified:	24 Jul 2023 01:45
URI:	http://repository.its.ac.id/id/eprint/98997

Actions (login required)

View Item