Optimasi Konsumsi Energi Dan Stabilitas Suhu Pada Alat Pengering Lada Menggunakan Metode Deep Reinforcement Learning (DRL)

Munandar, Alrifqi Berlin Ramadhan (2026) Optimasi Konsumsi Energi Dan Stabilitas Suhu Pada Alat Pengering Lada Menggunakan Metode Deep Reinforcement Learning (DRL). Bachelor thesis, Institut Teknologi Kalimantan.

[img] Text
04221021_cover.pdf
Restricted to Repository staff only until 4 October 2028.
Download (541kB) | Request a copy
[img] Text
04221021_statement_of_authenticity.pdf
Restricted to Repository staff only until 4 October 2028.
Download (366kB) | Request a copy
[img] Text
04221021_publishing_agreement.pdf
Restricted to Repository staff only until 4 October 2028.
Download (368kB) | Request a copy
[img] Text
04221021_preface.pdf
Restricted to Repository staff only until 4 October 2028.
Download (628kB) | Request a copy
[img] Text
04221021_abstract_id.pdf
Restricted to Repository staff only until 4 October 2028.
Download (643kB) | Request a copy
[img] Text
04221021_abstract_en.pdf
Restricted to Repository staff only until 4 October 2028.
Download (383kB) | Request a copy
[img] Text
04221021_table_of_content.pdf
Restricted to Repository staff only until 4 October 2028.
Download (370kB) | Request a copy
[img] Text
04221021_illustrations.pdf
Restricted to Repository staff only until 4 October 2028.
Download (345kB) | Request a copy
[img] Text
04221021_tables.pdf
Restricted to Repository staff only until 4 October 2028.
Download (336kB) | Request a copy
[img] Text
04221021_notations.pdf
Restricted to Repository staff only until 4 October 2028.
Download (504kB) | Request a copy
[img] Text
04221021_chapter_1.pdf
Restricted to Repository staff only until 4 October 2028.
Download (600kB) | Request a copy
[img] Text
04221021_chapter_2.pdf
Restricted to Repository staff only until 4 October 2028.
Download (746kB) | Request a copy
[img] Text
04221021_chapter_3.pdf
Restricted to Repository staff only until 4 October 2028.
Download (1MB) | Request a copy
[img] Text
04221021_chapter_4.pdf
Restricted to Repository staff only until 4 October 2028.
Download (4MB) | Request a copy
[img] Text
04221021_conclusions.pdf
Restricted to Repository staff only until 4 October 2028.
Download (458kB) | Request a copy
[img] Text
04221021_bibliography.pdf
Restricted to Repository staff only until 4 October 2028.
Download (389kB) | Request a copy
[img] Text
04221021_enclosure.pdf
Restricted to Repository staff only until 4 October 2028.
Download (1MB) | Request a copy
[img] Text
04221021_paper.pdf
Restricted to Repository staff only until 4 October 2028.
Download (463kB) | Request a copy
[img] Text
04221021_presentation.pdf
Restricted to Repository staff only until 4 October 2028.
Download (3MB) | Request a copy
[img] Text
04221021_Form. TA-020.pdf
Restricted to Repository staff only until 4 October 2028.
Download (425kB) | Request a copy
[img] Text
04221021_approval_sheet.pdf
Restricted to Repository staff only until 4 October 2028.
Download (409kB) | Request a copy
[img] Text
04221021_publishing_agreement.pdf
Restricted to Repository staff only until 4 October 2028.
Download (79kB) | Request a copy

Abstract

Kualitas lada kering (Piper nigrum L) sangat dipengaruhi oleh stabilitas suhu selama proses pascapanen. Sistem pengeringan konvensional dan sistem pengendalian berbasis aturan masih memiliki keterbatasan dalam menyesuaikan daya pemanasan terhadap perubahan kondisi termal, yang berpotensi menyebabkan ketidakstabilan suhu dan penggunaan energi yang kurang optimal. Penelitian ini mengembangkan sistem pengendalian suhu untuk pengering lada menggunakan metode Deep Reinforcement Learning (DRL) dengan algoritma Deep Q-Network (DQN) berbasis Digital Twin. Data empiris diperoleh dari prototipe pengering lada berbasis ESP32, dengan daya lampu pemanas sebagai masukan dan suhu ruang pengeringan sebagai keluaran sistem. Data ini digunakan dalam proses System Identification untuk membuat model Digital Twin berdasarkan nlarx1. Agen DQN dirancang dengan empat state, yaitu suhu ternormalisasi, error suhu, perubahan error, dan daya sebelumnya, serta tujuh aksi diskrit daya lampu dari 0 W hingga 360 W. Pengujian dilakukan dalam kondisi setpoint tracking, gangguan pengurangan daya, suhu awal tinggi, serta perbandingan dengan kontrol PID baseline. Hasil pengujian menunjukkan bahwa agen DQN mampu mencapai zona target 53–57°C pada detik ke-540, dengan suhu akhir 53,142°C, kesalahan akhir 1,858°C, MAE sebesar 2,048°C, RMSE sebesar 3,451°C, dan tanpa melampaui batas 60°C. Dalam skenario gangguan daya, agen tersebut mampu mengembalikan suhu ke zona target dalam waktu 90 detik setelah gangguan berhenti. Dibandingkan dengan PID dasar, DQN mencapai persentase waktu yang lebih tinggi di dalam zona target, yaitu 93,258% dibandingkan dengan 9,3633%. Oleh karena itu, DQN berbasis Digital Twin dapat dimanfaatkan sebagai strategi pengendalian suhu adaptif dalam sistem pengeringan lada.

Item Type: Thesis (Bachelor)
Subjects: S Agriculture > S Agriculture (General)
Divisions: Jurusan Teknologi Industri dan Proses > Teknik Elektro
Depositing User: Alrifqi Berlin Ramadhan Munandar
Date Deposited: 31 Jul 2026 05:46
Last Modified: 31 Jul 2026 05:46
URI: http://repository.itk.ac.id/id/eprint/27960

Actions (login required)

View Item View Item