Perancangan Algoritma Peramalan Daya Listrik di ITK Menggunakan Metode Long Short Term Memory

Devianti, Susan (2025) Perancangan Algoritma Peramalan Daya Listrik di ITK Menggunakan Metode Long Short Term Memory. Bachelor thesis, Institut Teknologi Kalimantan.

[img] Text
04211080_abstract_en.pdf
Restricted to Repository staff only until 5 October 2027.

Download (193kB) | Request a copy
[img] Text
04211080_abstract_id.pdf
Restricted to Repository staff only until 5 October 2027.

Download (255kB) | Request a copy
[img] Text
04211080_approval_sheet.pdf
Restricted to Repository staff only until 5 October 2027.

Download (561kB) | Request a copy
[img] Text
04211080_bibliography.pdf
Restricted to Repository staff only until 5 October 2027.

Download (249kB) | Request a copy
[img] Text
04211080_chapter_1.pdf
Restricted to Repository staff only until 5 October 2027.

Download (366kB) | Request a copy
[img] Text
04211080_chapter_2.pdf
Restricted to Repository staff only until 5 October 2027.

Download (654kB) | Request a copy
[img] Text
04211080_chapter_3.pdf
Restricted to Repository staff only until 5 October 2027.

Download (674kB) | Request a copy
[img] Text
04211080_chapter_4.pdf
Restricted to Repository staff only until 5 October 2027.

Download (2MB) | Request a copy
[img] Text
04211080_conclusions.pdf
Restricted to Repository staff only until 5 October 2027.

Download (255kB) | Request a copy
[img] Text
04211080_cover.pdf
Restricted to Repository staff only until 5 October 2027.

Download (187kB) | Request a copy
[img] Text
04211080_enclosure.pdf
Restricted to Repository staff only until 5 October 2027.

Download (1MB) | Request a copy
[img] Text
04211080_Form. TA-020.pdf
Restricted to Repository staff only until 5 October 2027.

Download (606kB) | Request a copy
[img] Text
04211080_illustrations.pdf
Restricted to Repository staff only until 5 October 2027.

Download (244kB) | Request a copy
[img] Text
04211080_notations.pdf
Restricted to Repository staff only until 5 October 2027.

Download (245kB) | Request a copy
[img] Text
04211080_paper.pdf
Restricted to Repository staff only until 5 October 2027.

Download (1MB) | Request a copy
[img] Text
04211080_preface.pdf
Restricted to Repository staff only until 5 October 2027.

Download (245kB) | Request a copy
[img] Text
04211080_presentation.pdf
Restricted to Repository staff only until 5 October 2027.

Download (2MB) | Request a copy
[img] Text
04211080_publishing_agreement.pdf
Restricted to Repository staff only until 5 October 2027.

Download (571kB) | Request a copy
[img] Text
04211080_statement_of_authenticity.pdf
Restricted to Repository staff only until 5 October 2027.

Download (549kB) | Request a copy
[img] Text
04211080_table_of_content.pdf
Restricted to Repository staff only until 5 October 2027.

Download (243kB) | Request a copy
[img] Text
04211080_tables.pdf
Restricted to Repository staff only until 5 October 2027.

Download (211kB) | Request a copy

Abstract

Energi listrik merupakan aspek penting dalam mendukung aktivitasi kehidupan sehari-hari, pada penelitian ini khususnya di lingkungan pendidikan seperti Gedung Laboratorium Institut Teknologi Kalimantan (ITK). Untuk meningkatkan kinerja optimal pada penggunaan daya listrik, diperlukan adanya sistem peramalan yang mampu memprediksi kebutuhan daya secara akurat. Penelitian ini berfokus pada perancangan serta penerapan model peramalan daya listrik dengan memanfaatkan pendekatan metode Long Short-Term Memory (LSTM). Data yang digunakan mencakup variabel multivariat seperti daya historis (P_Avg), suhu, sesi waktu, jam, dan hari, serta diuji dalam dua kondisi: data yang mengandung anomali dan data yang telah dibersihkan. Maka model dievaluasi menggunakan metrik RMSE, MAE, MSE, dan MAPE. Hasil terbaik diperoleh pada model LSTM dengan lima input dan data bersih, yang menghasilkan MAPE pengujian sebesar 16.82% serta kurva prediksi yang stabil dan mendekati tren aktual. Selain itu, model mampu mengenali beban puncak, yang umumnya terjadi pada siang hari dan hari kerja. Meskipun model GRU memberikan MAPE lebih rendah pada konfigurasi tertentu (15.77%), model LSTM tetap dinilai lebih unggul secara keseluruhan dalam hal akurasi. Sementara itu, model RNN menunjukkan performa terendah karena keterbatasan dalam mengatur prediksi jangka panjang. Hasil ini menunjukkan bahwa LSTM merupakan algoritma yang paling tepat untuk digunakan dalam sistem peramalan daya listrik di lingkungan kampus. Kata kunci: Evaluation Error, LSTM, Peramalan Daya Listrik

Item Type: Thesis (Bachelor)
Subjects: T Technology > T Technology (General)
Divisions: Jurusan Teknologi Industri dan Proses > Teknik Elektro
Depositing User: Susan Devianti
Date Deposited: 11 Jul 2025 00:31
Last Modified: 11 Jul 2025 00:31
URI: http://repository.itk.ac.id/id/eprint/24015

Actions (login required)

View Item View Item