Rancang Bangun Software Predictive Maintenance Berbasis Kecerdasan Buatan untuk Transformator

Halim, Ardiyanto (2023) Rancang Bangun Software Predictive Maintenance Berbasis Kecerdasan Buatan untuk Transformator. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07111940000097-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (4MB) | Request a copy

Abstract

Maintenance/perawatan merupakan salah satu hal yang perlu diperhatikan pada operasional transformator, karena cukup kompleks dan biaya yang dikeluarkan tidak sedikit. maka dari itu pemanfaatan predictive maintenance sangat dibutuhkan dalam penggunaan transformator dimana maintenance jenis ini dapat memperkirakan kapan seharusnya dilaksanakan maintenance agar lebih efisien. Namun, untuk pelaksanaan predictive maintenance sendiri memiliki proses yang cukup rumit dimana berbagai aspek pengujian dan analisa parameter seperti Dissolve Gas Analysis (DGA), Breakdown Voltage (BDV) dan Water Content sehingga membutuhkan tenaga ahli untuk melakukan interpretasi atau analisa terhadap parameter parameter tersebut. Maka untuk meningkatkan efisiensi dan keandalan dari operasional transformator, pada tugas akhir kali ini dilakukan pembuatan software predictive maintena ce berbasis kecerdasan buatan (Artificial Intelligence) dengan bahasa pemograman Python yang dapat melakukan interpretasi parameter DGA, BDV dan Water Content. hasil dari interpretasi parameter-parameter tersebut akan melalui proses masing-masing dan akan menghasilkan keluaran berupa status BDV, Status Water Content, status oli, Jenis Fault (bila ada), kondisi kertas insulasi, probabilitas fault, interval sampling, rekomendasi tindakan dan grafik untuk monitoring. digunakan berbagai metode seperti Machine Learning untuk menghasilkan keluaran jenis fault, Fuzzy Logic untuk keluaran kondisi kertas insulasi, rulebased reasoning untuk keluaran rekomendasi tindakan dan keluaran lainnya berdasarkan standar internasional seperti IEEE C57.104 dan IEC 60599. Pembuatan keluaran jenis fault berbasis machine learning juga dievaluasi lebih lanjut dengan hasil akurasi yang 11,93% lebih tinggi dibandingkan metode konvensional pada data yang sama. Dengan menggunakan berbagai metode kecerdasan buatan, software ini mampu menganalisis data parameter masukan dan mendeteksi potensi masalah atau fault sebelum terjadinya kegagalan. Hal ini dapat menghindari kerusakan trafo yang tak terduga, mengurangi downtime, dan meningkatkan keandalan sistem kelistrikan.
====================================================================================================================================
Maintenance is one of the things that needs to be considered in transformer operations, because it is quite complex and the costs incurred are not small. Therefore, the use of predictive maintenance is needed in the use of transformers where this type of maintenance can predict when maintenance should be carried out to make it more efficient. However, the implementation of predictive maintenance itself has a complicated process where various aspects of testing and parameter analysis such as Dissolve Gas Analysis (DGA), Breakdown Voltage (BDV) and Water Content that require experts to interpret or analyze these parameters. So, to increase the efficiency and reliability of transformer operations, in this final project, an artificial intelligence (AI) based predictive maintenance software is developed using the Python programming language which can interpret DGA, BDV and Water Content parameters. the results of the interpretation of these parameters will go through their respective processes and will produce output in the form of BDV status, Water Content status, oil status, fault type (if any), insulation paper condition, fault probability, sampling interval, action recommendations and graphs for monitoring. various methods are used such as Machine Learning to produce fault type output, Fuzzy Logic to output insulation paper conditions, rule-based reasoning to output action recommendations and other outputs based on international standards such as IEEE C57.104 and IEC 60599. Machine learning based fault type output generation also evaluated further with results of accuracy which is 11.93% higher than conventional methods on the same data. By using various AI methods, this software can analyze input parameter data and detect potential problems or faults before failure occurs. This can avoid unforeseen transformer breakdowns, reduce downtime, and increase the reliability of the electrical system.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Artificial Intelligence, Dissolve Gas Analysis, Breakdown Voltage, Water Content
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Q Science > QA Mathematics > QA336 Artificial Intelligence T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK6565.T7 Transformers
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Ardiyanto Halim
Date Deposited:	22 Jul 2023 13:27
Last Modified:	22 Jul 2023 13:27
URI:	http://repository.its.ac.id/id/eprint/98772

Actions (login required)

View Item