Firlana, Fakihatu Abdi (2023) Perancangan Sistem Deteksi Kesalahan Winding Short Motor Induksi Berbasis Model Estimasi Long Short-Term Memory (LSTM) pada Electric Submersible Pump (ESP) di Sumur Minyak UPB-12 PT. SIPL Gresik. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 02311940000069-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
02311940000069-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (4MB) | Request a copy |
Abstract
Salah satu kesalahan yang sering terjadi pada Electric Submersible Pump (ESP) adalah kesalahan pada motor induksi. Sebanyak 16% statistik kesalahan motor induksi disebabkan karena adanya hubungan pendek (short-circuit) pada stator winding. Apabila tidak ada sistem deteksi dini karena fungsinya yang vital, maka akan menyebabkan kerusakan fatal. Penelitian ini bertujuan merancang model estimasi menggunakan LSTM dan residual analysis untuk mendeteksi kesalahan winding short motor induksi dengan algoritma klasifikasi. Data variabel motor induksi yang digunakan meliputi tegangan listrik tiga fasa, temperatur, dan arus listrik. Terdapat dua model estimasi yang dirancang, yaitu Model A menggunakan data tegangan listrik tiga fasa dan data temperatur serta Model B menggunakan data tegangan listrik tiga fasa dan data arus listrik. Nilai nRMSE Model A didapatkan sebesar 0,523% dan nilai nRMSE Model B didapatkan sebesar 1,555%. Kemudian dilakukan pembangkitan residu dan ekstraksi fitur yang menghasilkan delapan fitur statistik dengan kategori healthy, warning, dan fault. Hasil ekstraksi fitur yang didapatkan akan dilakukan feature ranking dan feature selection menggunakan algoritma ReliefF. Variasi fitur digunakan sebagai perbandingan untuk memilih algoritma classifer terbaik. Classifier yang digunakan yaitu Decision Tree, Gaussian Naïve Bayes, SVM, k-NN, dan Bagged Trees. Classifier terbaik Model A dihasilkan algoritma Fine k-NN dengan akurasi 98,214% yang menggunakan 1 fitur. Classifier terbaik Model B dihasilkan algoritma Decision Tree dengan akurasi 98,246% yang menggunakan 1 fitur. Classifier terbaik gabungan Model A dan Model B dihasilkan algoritma Decision Tree dengan akurasi 94,643% yang menggunakan 1 fitur.
=======================================================================================================================================
One of the frequent faults that commonly occur in an Electric Submersible Pump (ESP) is a fault in the induction motor. Approximately 16% of induction motor faults are caused by short circuits in the stator winding. The absence of an early detection system, due to its crucial function, can result in significant damage. The objective of this study is to develop an estimation model using the LSTM algorithm and residual analysis to identify faults related to winding short circuits in the induction motor using a classification algorithm. The data variables utilized consist of three-phase electrical voltage, temperature, and electrical current. Two estimation models were created, namely Model A, which employs three-phase electrical voltage and temperature, and Model B, which uses three-phase electrical voltage and electrical current. The nRMSE value obtained for Model A is 0.523%, while for Model B it is 1.555%. Subsequently, residue generation and extraction of statistical features were performed, resulting in eight features categorized as healthy, warning, and fault. The extracted features will undergo feature ranking using the ReliefF algorithm. Features variation will be employed for comparison in order to select the classifier algorithm with the best performance. The classifiers utilized include Decision Tree, Gaussian Naïve Bayes, SVM, k-NN, and Bagged Trees. The best-performing classifier for Model A is the Fine k-NN, which achieves 98.814% accuracy and utilizes 1 feature. The best classifier for Model B is the Decision Tree, which achieves 98.246% accuracy and utilizes 1 feature. The best-combined classifier for Model A and Model B is the Decision Tree, which achieves 94.643% accuracy and utilizes 1 feature.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Classifier, LSTM, Model Estimasi, Motor Induksi, Residu, Classifier, Estimation Model, Induction Motor, LSTM, Residue |
| Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > T Technology (General) > T57.5 Data Processing T Technology > TJ Mechanical engineering and machinery > TJ910 Electric pumping machinery T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK2785 Electric motors, Induction. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7870.23 Reliability. Failures |
| Divisions: | Faculty of Industrial Technology > Physics Engineering > 30201-(S1) Undergraduate Thesis |
| Depositing User: | Fakihatu Abdi Firlana |
| Date Deposited: | 27 Jul 2023 03:08 |
| Last Modified: | 27 Jul 2023 03:10 |
| URI: | http://repository.its.ac.id/id/eprint/99316 |
Actions (login required)
 |
View Item |