Deteksi Kondisi Vibrasi Forced Draft Fan di Unit Package Boiler PT. XYZ dengan Metode Long Short Term Memory (LSTM)

Oktamutia, Tsabitha Kanaya (2023) Deteksi Kondisi Vibrasi Forced Draft Fan di Unit Package Boiler PT. XYZ dengan Metode Long Short Term Memory (LSTM). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 02311940000061-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (4MB) | Request a copy

Abstract

Rotating equipment memiliki peranan penting dalam berbagai sektor industri. Pada faktanya, banyak penggunaan kontinyu dari mesin industri seperti pompa dan turbin gas bergantung pada kinerja rotating equipment-nya. Vibrasi atau getaran sering kali ditemui pada berbagai kasus yang melibatkan mesin yang berputar atau rotating machinery. Salah satu rotating equipment esensial pada sebuah boiler adalah Forced Draft Fan atau dapat disebut sebagai FD Fan. Pada penelitian ini penulis mencoba melakukan prediksi vibrasi pada vibrasi FD Fan menggunakan neural networks metode Long Short Term Memory (LSTM). Data yang digunakan untuk pengolahan pada model neural networks berupa data historis preventive maintenance pada PT. XYZ berupa data kecepatan vibrasi selama tiga tahun terakhir. Secara spesifiknya data yang diolah yaitu data vibrasi pada rotating equipment berupa FD Fan (01-GB-0323) dengan posisi vertikal dan horizontal pada bagian Drive End dan Non-Drive End secara time series. Pengolahan data dilakukan dengan dua tahap yaitu analisis statistik data vibrasi dan perancangan model prediksi vibrasi dengan metode LSTM. Dari hasil analisis statistik data vibrasi, didapatkan data vibrasi 1 atau Non Drive End cenderung memiliki standar deviasi yang lebih tinggi karena berada dekat dengan posisi rumah bearing. LSTM berhasil memprediksi nilai vibrasi beberapa bulan ke depan dengan dilakukan perhitungan Root Mean Square Error (RMSE). Didapatkan nilai rata-rata RMSE pada Vibrasi 1 NDE Vertikal 0.11 dan 0.09, Vibrasi 2 DE Vertikal 0.15 dan 0.15, Vibrasi 1 NDE Horizontal 0.15 dan 0.15, serta Vibrasi 2 DE Horizontal 0.15 dan 0.15 masing-masing terhitung pada nilai train score serta test score. Hasil model prediksi yang dibangun dengan menggunakan metode LSTM memiliki nilai pengujian rata-rata RMSE < 1 atau mendekati 0 sehingga dapat dikatakan sangat baik.
================================================================================================================================
Rotating equipment has an important role in various industrial sectors. In fact, many of the continuous uses of industrial machines such as pumps and gas turbines depend on the performance of the rotating equipment. Vibration is often encountered in various cases involving rotating machines. One of the essential rotating equipment in a boiler is a Forced Draft Fan or can be referred to as an FD Fan. In this study the authors tried to predict vibration in the vibration of the FD Fan using the Long Short Term Memory (LSTM) neural networks method. The data used for processing the neural networks model is in the form of historical preventive maintenance data at PT. XYZ is the vibration velocity data for the last three years. Specifically, the data processed is vibration data on rotating equipment in the form of an FD Fan (01-GB-0323) with vertical and horizontal positions on the Drive End and Non-Drive End in time series. Data processing is carried out in two stages, namely statistical analysis of vibration data and design of prediction models using the LSTM method. From the results of statistical analysis of vibration data, it is obtained that vibration data 1 or Non Drive End tends to have a higher standard deviation because it is close to the position of the bearing housing. LSTM succeeded in predicting the vibration value for the next few months by calculating the Root Mean Square Error (RMSE). The average RMSE values were obtained for Vibration 1 NDE Vertical 0.11 and 0.09, Vibration 2 DE Vertical 0.15 and 0.15, Vibration 1 NDE Horizontal 0.15 and 0.15, and Vibration 2 DE Horizontal 0.15 and 0.15 respectively calculated on the value of train score and test score .The results of the prediction model built using the LSTM method have an average test value of RMSE <1 or close to 0 so that it can be said to be very good.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deteksi Kondisi, FD Fan, LSTM, Predictive Maintenance, Vibrasi
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > T Technology (General) T Technology > T Technology (General) > T174 Technological forecasting T Technology > TJ Mechanical engineering and machinery > TJ174 Maintenance and repair of machinery
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Tsabitha Kanaya Oktamutia
Date Deposited:	07 Aug 2023 05:27
Last Modified:	07 Aug 2023 05:27
URI:	http://repository.its.ac.id/id/eprint/103648

Actions (login required)

View Item