Saraswati, Luh Putu Intan (2026) Analisis Model Deteksi Kelelahan Kerja berbasis Machine Learning: Studi Kasus pada Tim Pemeliharaan Pembangkit Listrik Bali. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of ACCESS CLOSED BY AUTHOR]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text (ACCESS CLOSED BY AUTHOR)
5010221059-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy |
|
Text
5010221059-Surat Permohonan Full Closed Access.pdf - Accepted Version Restricted to Repository staff only Download (173kB) | Request a copy |
Abstract
Kelelahan kerja merupakan faktor risiko yang dapat menurunkan kewaspadaan, kecepatan respons, dan performa pekerja, terutama pada pekerjaan yang menuntut kesiapan fisik dan mental secara simultan, seperti aktivitas tim pemeliharaan pada sektor pembangkit listrik. Namun, perusahaan pembangkit listrik Bali belum memiliki pengukuran kelelahan kerja berbasis model yang terstruktur untuk mendukung pengendalian risiko di lapangan. Penelitian ini bertujuan menganalisis fitur Heart Rate Variability (HRV) yang paling relevan dalam merepresentasikan kelelahan kerja, menerapkan model deteksi kelelahan berbasis machine learning, serta memberikan rekomendasi pengendalian kelelahan kerja bagi perusahaan. Penelitian ini terdiri dari 9 tahapan, yakni persiapan, studi pendahuluan, perencanaan, pengumpulan data, pre-processing data, uji signifikansi fitur HRV, analisis penerapan model, penerapan model, dan pemberian rekomendasi. Pada pengumpulan data, penelitian ini memperoleh 23 responden dari tim pemeliharaan melalui pengukuran sebelum dan sesudah bekerja sehingga menghasilkan 46 data observasi. HRV berupa Mean RR, RMSSD, dan LF/HF Ratio digunakan sebagai indikator objektif respons fisiologis tubuh, sedangkan Rating of Perceived Exertion (RPE) digunakan sebagai dasar pelabelan kondisi kelelahan kerja. Hasil penelitian menunjukkan bahwa LF/HF Ratio merupakan fitur yang paling mampu merepresentasikan kelelahan kerja karena menunjukkan perbedaan yang signifikan antara kelompok No Fatigue dan Mild Fatigue dengan nilai p-value sebesar 0,001. Sedangkan, Mean RR dan RMSSD tidak menunjukkan perbedaan yang signifikan dengan p-value masing-masing sebesar 0,367 dan 0,388. Model deteksi kelelahan diterapkan menggunakan algoritma Support Vector Machine (SVM) dengan empat jenis kernel, yakni polynomial, linear, radial basis function (RBF), dan sigmoid. Hasil evaluasi menunjukkan bahwa SVM Polynomial menghasilkan kinerja terbaik dengan akurasi sebesar 87%, F1-score sebesar 0,875 pada label No Fatigue, dan 0,864 pada label Mild Fatigue. Hasil Confusion Matrix juga menunjukkan jumlah kesalahan klasifikasi yang relatif rendah, yaitu 2 data false positive atau false alarm dan 4 data false negative. Berdasarkan hasil penerapan model, perusahaan direkomendasikan melakukan pemantauan kelelahan secara rutin dan objektif menggunakan device yang tersedia serta menerapkan administrative control berupa SOP, microbreak dan rotasi tugas untuk mengurangi risiko human error akibat kelelahan.
=====================================================================================================================================
Work fatigue is a risk factor that can reduce worker alertness, response speed, and performance, especially in jobs that demand simultaneous physical and mental readiness, such as the activities of maintenance teams in the power plant sector. However, the Bali power plant company does not yet have a structured model-based work fatigue measurement to support risk control in the field. This study aims to analyze the most relevant Heart Rate Variability (HRV) features in representing work fatigue, implement a machine learning-based fatigue detection model, and provide recommendations for work fatigue control for the company. This study consists of 9 stages, namely preparation, preliminary study, planning, data collection, data pre-processing, HRV feature significance testing, model application analysis, model application, and providing recommendations. In data collection, this study obtained 23 respondents from the maintenance team through pre- and post-work measurements, resulting in 46 observation data. HRV in the form of Mean RR, RMSSD, and LF/HF Ratio were used as objective indicators of the body's physiological response, while the Rating of Perceived Exertion (RPE) was used as the basis for labelling work fatigue conditions. The results showed that the LF/HF Ratio was the most capable feature representing work fatigue because it showed a significant difference between the No Fatigue and Mild Fatigue groups with a p-value of 0.001. Meanwhile, the Mean RR and RMSSD did not show a significant difference with p-values of 0.367 and 0.388, respectively. The fatigue detection model was implemented using the Support Vector Machine (SVM) algorithm with four types of kernels, namely polynomial, linear, radial basis function (RBF), and sigmoid. The evaluation results showed that the Polynomial SVM produced the best performance with an accuracy of 87%, an F1-score of 0.875 for the No Fatigue label, and 0.864 for the Mild Fatigue label. The Confusion Matrix results also showed a relatively low number of classification errors, namely 2 false positive or false alarm data and 4 false negative data. Based on the results of the model implementation, companies are recommended to conduct routine and objective fatigue monitoring using available devices and implement administrative controls in the form of SOPs, microbreaks and task rotation to reduce the risk of human error due to fatigue.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Kelelahan Kerja, Heart Rate Variability, Rating of Perceived Exertion, Machine Learning, Support Vector Machine. |
| Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > T Technology (General) > T55 Industrial Safety |
| Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26201-(S1) Undergraduate Thesis |
| Depositing User: | Luh Putu Intan Saraswati |
| Date Deposited: | 27 Jul 2026 01:27 |
| Last Modified: | 27 Jul 2026 01:34 |
| URI: | http://repository.its.ac.id/id/eprint/137929 |
Actions (login required)
 |
View Item |