Perancangan Sistem Aplikasi Prediksi Risiko pada Koneksi Komponen Listrik Panel Control Motor Chiller Menggunakan Thermal Imagers Fluke dengan Metode AI-Vision

Abstract

The high demand for data and information in the digital era requires the availability of supporting facilities in the form of data centers. As a physical facility, a data center consists of an array of interconnected electronic devices that form a system. Large amounts of data processing activity can cause electronic devices to become overheated. Overheated temperatures on electronic components can cause problems such as damage to devices and decreased performance, resulting in the loss of important data and threatening the continuity of business operations. Optimal data center temperatures can be maintained by using a chiller that is monitored regularly. Through this research, the authors designed the AI-Vision system to make risk predictions for chillers in the H2-01 Karawang data center. AI-Vision is an artificial intelligence application system using machine learning methods to calculate the risk prediction of the condition of electrical components in a chiller based on images captured by thermal imagers fluke. The AI-Vision application uses image processing and the K-Nearest Neighbors algorithm to produce risk predictions that are adjusted to standards issued by PT. DCI by adopting IEC 60269 and PUIL 2011 standards, as well as research conducted by EPRI (Electric Power Research Institute). The results of this study are in the form of a risk prediction with an accuracy value of 90%.