Implementasi Algoritma K-Nearest Neighbors Untuk Klasifikasi Kecepatan Getaran Pada Rotor Kit Bently Nevada RK-4

Rusdi, Alfin Alhimni (2024) Implementasi Algoritma K-Nearest Neighbors Untuk Klasifikasi Kecepatan Getaran Pada Rotor Kit Bently Nevada RK-4. Diploma thesis, Institut Teknologi Sepuluh Nopember.

Text 2040201049-Undergraduate_Theses.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (5MB) | Request a copy

Abstract

Rotating equipment merupakan peralatan vital dalam proses-proses di pembangkit listrik, seperti dalam sistem aliran udara dan gas buang pada PLTU terdapat ID Fan. Pada tanggal 18 November 2021, ID Fan PLTU yang dikelola oleh PT PLN Nusantara Power Services mengalami permasalahan unbalance dan mechanical looseness akibat batubara yang dihisap oleh ID Fan mengalami penumpukan dan penggumpalan pada impeller, hal ini menyebabkan penambahan massa pada impeller ID Fan. Semakin menumpuknya batubara pada impeller dapat mengakibatkan gesekan antara impeller dengan shaft rotor sehingga terjadi peningkatan getaran, yang dampaknya bisa menyebabkan impeller patah karena ketidakseimbangan massa serta kelonggaran baut pada bearing. Sehingga dilakukan perawatan pada ID Fan dengan cara penjadwalan rutin untuk menjaga nilai batas aman getaran sesuai dengan standar ISO 10816. Perawatan berdasarkan penjadwalan rutin memiliki beberapa keterbatasan yaitu ketidakmampuannya untuk mengetahui kondisi aktual ID Fan di antara interval waktu penjadwalan rutin yang telah ditentukan. Kerusakan yang terjadi pada ID Fan di pembangkit dapat menyebabkan unit trip dan derating. Oleh sebab itu berdasarkan permasalahan yang ada di power plan, dibuat alat inovasi sistem monitoring getaran yang dapat mengklasifikasikan kondisi aktual dari ID Fan untuk membantu tim operation dan pemeliharaan dalam merencanakan perawatan dan perbaikan. Sistem monitoring getaran ini mengimplementasikan klasifikasi kondisi rotor kit. Rotor kit digunakan untuk mensimulasikan kerusakan unbalance dan mechanical looseness pada ID Fan. Klasifikasi kondisi rotor kit menggunakan metode K-Nearest Neighbors yang merupakan sebuah metode klasifikasi berdasarkan jarak tetangga terdekat. Sinyal kecepatan getaran direkam menggunakan sensor adxl345 dan diolah menggunakan Fast Fourier Transform, sehingga didapatkan sinyal kecepatan getaran dalam domain frekuensi rendah maupun frekuensi tinggi. Dari penelitian yang telah dilakukan, arsitektur model K-Nearest Neighbors terbaik menggunakan pembagian data training sebesar 80% dan data testing sebesar 20% dengan parameter nilai k= 9 didapatkan nilai akurasi sebesar 93%, presisi sebesar 94%, recall sebesar 93%, dan f1-score sebesar 93%.
===================================================================================================================================
Rotating equipment is vital equipment in the processes in power plants, such as in the air and exhaust gas flow system at the PLTU there is an ID Fan. On November 18, 2021, the ID Fan of the PLTU managed by PT PLN Nusantara Power Services experienced unbalance and mechanical looseness problems due to coal sucked by the ID Fan accumulating and clumping on the impeller, this caused an increase in mass on the ID Fan impeller. The increasing accumulation of coal on the impeller can cause friction between the impeller and the rotor shaft, resulting in increased vibration, the impact of which can cause the impeller to break due to mass imbalance and loose bolts on the bearing. So that maintenance is carried out on the ID Fan by means of routine scheduling to maintain the safe vibration limit value according to the ISO 10816 standard. Maintenance based on routine scheduling has several limitations, namely its inability to know the actual condition of the ID Fan between the specified routine scheduling time intervals. Damage to the ID Fan at the power plant can cause unit trips and derating. Therefore, based on the problems in the power plan, an innovative vibration monitoring system tool was created that can classify the actual condition of the ID Fan to help the operation and maintenance team in planning maintenance and repairs. This vibration monitoring system implements the rotor kit condition classification. The rotor kit is used to simulate unbalance and mechanical looseness damage to the ID Fan. The rotor kit condition classification uses the K-Nearest Neighbors method which is a classification method based on the nearest neighbor distance. The vibration velocity signal is recorded using the adxl345 sensor and processed using Fast Fourier Transform, so that the vibration velocity signal is obtained in the low frequency and high frequency domains. From the research that has been done, the best K-Nearest Neighbors model architecture uses a training data division of 80% and testing data of 20% with a parameter value of k = 9, an accuracy value of 93%, a precision of 94%, a recall of 93%, and an f1-score of 93%.

Item Type:	Thesis (Diploma)
Uncontrolled Keywords:	Fast-Fourier Transform, Getaran, ID Fan, K-Nearest Neighbors, Mechanical Looseness, Rotating Equipment, Unbalance.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Alfin Alhimni Rusdi
Date Deposited:	09 Sep 2024 01:05
Last Modified:	09 Sep 2024 01:05
URI:	http://repository.its.ac.id/id/eprint/115607

Actions (login required)

View Item