SISTEM MONITORING DAN DETEKSI ANOMALI GETARAN DAN TEMPERATUR MOTOR 3 PHASE MENGGUNAKAN METODE K-MEANS CLUSTERING BERBASIS EDGE AI

Fahrul Ridho Dwinugroho, . (2026) SISTEM MONITORING DAN DETEKSI ANOMALI GETARAN DAN TEMPERATUR MOTOR 3 PHASE MENGGUNAKAN METODE K-MEANS CLUSTERING BERBASIS EDGE AI. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

[img] Text
COVER.pdf

Download (1MB)
[img] Text
BAB 1.pdf

Download (280kB)
[img] Text
BAB 2.pdf
Restricted to Registered users only

Download (953kB) | Request a copy
[img] Text
BAB 3.pdf
Restricted to Registered users only

Download (681kB) | Request a copy
[img] Text
BAB 4.pdf
Restricted to Registered users only

Download (871kB) | Request a copy
[img] Text
BAB 5.pdf
Restricted to Registered users only

Download (358kB) | Request a copy
[img] Text
DAFTAR PUSTAKA.pdf

Download (219kB)
[img] Text
LAMPIRAN .pdf
Restricted to Registered users only

Download (445kB) | Request a copy

Abstract

Motor induksi tiga fasa merupakan penggerak utama dalam industri yang rentan terhadap penurunan performa akibat masalah mekanis dan termal. Pemantauan kondisi mesin secara konvensional menggunakan perangkat otomasi standar seringkali membutuhkan biaya investasi yang mahal, sedangkan sistem berbasis Cloud atau Internet of Things (IoT) kerap terkendala oleh masalah latensi jaringan, tingginya konsumsi bandwidth, dan risiko keamanan privasi data. Penelitian ini bertujuan untuk merancang bangun purwarupa sistem pemantauan mandiri (stand-alone) menggunakan pendekatan Edge Artificial Intelligence (Edge AI) berbasis Tiny Machine Learning (TinyML) yang diimplementasikan pada mikrokontroler ESP32, terintegrasi dengan modul Ethernet W5500 melalui protokol Modbus TCP tanpa ketergantungan pada cloud. Sistem ini mengintegrasikan sensor MPU6050 untuk membaca kecepatan getaran (Velocity RMS) dan sensor termokopel untuk membaca suhu permukaan motor secara real-time. Sebagai metode klasifikasi, digunakan algoritma unsupervised learning K-Means Clustering, dengan jumlah cluster optimal (K = 2) ditentukan melalui evaluasi Silhouette Score dan Davies-Bouldin Index, menghasilkan dua kategori kondisi motor yaitu Normal dan Anomali (Danger), dengan interpretasi ambang batas mengacu pada standar industri ISO 10816 dan deteksi suhu berlebih (Overheating). Hasil pengujian menunjukkan model klasifikasi telah diimplementasikan pada perangkat edge dengan tingkat kesesuaian cluster sebesar 99,19% dan konsistensi deteksi anomali sebesar 94,7–98,2% terhadap spesifikasi model saat divalidasi pada pengujian operasional selama kurang lebih 8 jam pada motor uji. Sistem yang dihasilkan mampu melakukan proses inferensi secara langsung di perangkat (on-device inference) dengan latensi rendah tanpa memerlukan koneksi internet publik, sehingga dapat memberikan solusi peringatan dini (early warning) pemeliharaan prediktif yang efisien, aman, dan berbiaya rendah bagi industri. Kata Kunci: Motor Induksi, Edge AI, TinyML, K-Means Clustering, ESP32, ISO 10816, Deteksi Anomali, Pemeliharaan Prediktif.*****Three-phase induction motors are the primary drivers in industry, highly vulnerable to performance degradation due to mechanical and thermal issues. Conventional machine condition monitoring using standard automation devices often requires high investment costs, while Cloud or Internet of Things (IoT)-based systems frequently face challenges such as network latency, high bandwidth consumption, and data privacy security risks. This study aims to design and develop a stand-alone monitoring system prototype using an Edge Artificial Intelligence (Edge AI) approach based on Tiny Machine Learning (TinyML) implemented on an ESP32 microcontroller, integrated with a W5500 Ethernet module via the Modbus TCP protocol without dependence on the cloud. The system integrates an MPU6050 sensor to measure vibration velocity (Velocity RMS) and a thermocouple sensor to monitor the motor's surface temperature in real-time. As the classification method, an unsupervised K-Means Clustering algorithm was used, with the optimal number of clusters (K = 2) determined through Silhouette Score and Davies-Bouldin Index evaluation, resulting in two motor condition categories: Normal and Anomaly (Danger), with threshold interpretation referring to the ISO 10816 industrial standard and overheating detection. Test results show that the classification model, once implemented on the edge device, achieved a cluster-assignment consistency of 99.19% and an anomaly-detection consistency of 94.7–98.2% against the model specification when validated through approximately 8 hours of operational testing on a test motor. The resulting system is capable of performing on-device inference with low latency without requiring a public internet connection, thereby offering an efficient, secure, and low-cost predictive maintenance early-warning solution for industry. Keywords: Induction Motor, Edge AI, TinyML, K-Means Clustering, ESP32, ISO 10816, Anomaly Detection, Predictive Maintenance.

Item Type: Thesis (Sarjana)
Additional Information: 1.) Ir. Heri Firmansyah, M.T 2.) Rizki Pratama Putra, S.T.,M.T
Subjects: Teknologi dan Ilmu Terapan > Teknik Elektronika
Teknologi dan Ilmu Terapan > Teknik Komputer
Divisions: FT > D IV Teknologi Rekayasa Otomasi
Depositing User: Fahrul Ridho Dwinugroho .
Date Deposited: 07 Aug 2026 03:15
Last Modified: 07 Aug 2026 03:15
URI: http://repository.unj.ac.id/id/eprint/68718

Actions (login required)

View Item View Item