Estimasi Remaining Useful Life Untuk Keperluan Predictive Maintenance Menggunakan Model Degradasi Pada Hydrogenerator

Hartati, Ayu Dian (2024) Estimasi Remaining Useful Life Untuk Keperluan Predictive Maintenance Menggunakan Model Degradasi Pada Hydrogenerator. Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 6009221007-Master_Thesis.pdf] Text
6009221007-Master_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2026.

Download (2MB) | Request a copy

Abstract

Pembangkit listrik tenaga air mengalami perkembangan signifikan karena adanya kebutuhan pemanfaatan energi alternatif. Hydro-turbine generator (HGU) adalah komponen yang memerlukan perhatian khusus dalam pemeliharaan. Permasalahan generator umumnya berkaitan dengan adanya air-gap eccentricity. Remaining useful life (RUL) generator dapat diketahui dengan melakukan pemodelan degradasi eksponensial menggunakan predictive maintenance. Pengumpulan data dilakukan melalui simulasi synchronous machine dengan kondisi operasional sebesar 15 MW serta penambahan resistansi sebagai bentuk kesalahan sistem yang mewakilkan periode waktu 61 hari. Penelitian ini juga menggunakan metode Time-Domain Feature Extraction, Monotonicity, Principal Component Analysis (PCA) dan Health Indicator (HI) dengan penggunaan nilai monotonisitas yang lebih besar dari 0,7. Estimasi RUL pada hari ke 40, PDF menampilkan bahwa dengan probabilitas kepercayaan 0.15, sisa masa pakai sistem kurang lebih 25 hari. estimasi RUL mencapai ambang batas (threshold) yakni, 155,35 atau saat health indicator berhenti (akhir). Nilai garis dari estimasi RUL menampilkan bahwa sisa masa pakai sistem tersisa 0 hari, begitupun dengan nilai garis true RUL, probabilitas kepercayaan pada hari ke 60 ini adalah 0.6. Probabilitas dengan variasi α 20% meningkat secara signifikan, mencapai puncak mendekati nilai 0.9 sekitar hari ke-30 hingga hari ke-50. Model RUL yang digunakan ini menghasilkan nilai RMSE 0,924.
========================================================================================================================
Hydroelectric power plants have experienced significant development due to the need for alternative energy utilization. Hydro-turbine generators (HGU) are components that require special maintenance attention. Generator problems are commonly related to the presence of air-gap eccentricity. The remaining useful life (RUL) of the generator can be determined by modelling exponential degradation using predictive maintenance. Data collection was carried out through synchronous machine simulation with operational conditions of 15 MW and the addition of resistance as a form of system error representing a period of 61 days. This research also uses the Time-Domain Feature Extraction, Monotonicity, Principal Component Analysis (PCA) and Health Indicator (HI) methods using monotonicity values greater than 0.7. The RUL estimation on day 40, PDF displays that with a confidence probability of 0.15, the remaining life of the system is approximately 25 days. The RUL estimation reaches the threshold, namely, 155.35 or when the health indicator stops (end). The line value of the estimated RUL displays that the remaining system lifetime is 0 days, as well as the true RUL line value, the confidence probability at day 60 is 0.6. The probability with a 20% variation of α increases significantly, reaching a peak near the value of 0.9 around day 30 to day 50. The RUL model used produces an RMSE value of 0.924.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Remaining Useful Life, Model Degradasi Eksponensial, Synchronous Machine, Air -gap eccentricity. Remaining Useful Life, Exponential Degradation Model, Synchronous Machine, Air -gap eccentricity
Subjects: T Technology > T Technology (General) > T57.5 Data Processing
Divisions: Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30101-(S2) Master Thesis
Depositing User: Ayu Dian Hartati
Date Deposited: 11 Aug 2024 12:18
Last Modified: 11 Aug 2024 12:18
URI: http://repository.its.ac.id/id/eprint/114522

Actions (login required)

View Item View Item