Prediksi Energi Keluaran PLTS Off-Grid Berbasis Jaringan Syaraf Tiruan Backpropagation
DOI:
https://doi.org/10.36312/2zvn8v57Keywords:
Plts Off-Grid, Energi Keluaran, Peramalan, Mape, Backpropagation Neural.Abstract
Pembangkit Listrik Tenaga Surya (PLTS) off-grid memiliki tantangan dalam menjaga ketersediaan energi akibat fluktuasi produksi yang dipengaruhi oleh kondisi cuaca. Oleh karena itu, diperlukan metode prediksi energi yang dapat membantu memperkirakan ketersediaan energi pada periode berikutnya. Penelitian ini bertujuan mengembangkan model prediksi energi keluaran PLTS off-grid menggunakan metode Backpropagation Feed Forward Neural Network. Data yang digunakan berupa energi keluaran PLTS, suhu lingkungan, dan iradiasi matahari yang dikumpulkan dari PLTS off-grid di IT-PLN selama tiga hari berturut-turut dengan interval pengukuran 15 menit pada rentang waktu pukul 09.00–15.00 WIB. Model jaringan syaraf tiruan dibangun menggunakan 75 neuron input yang terdiri atas data energi, suhu, dan iradiasi, satu hidden layer, serta 25 neuron output. Penentuan jumlah neuron pada hidden layer dan nilai learning rate dilakukan menggunakan metode trial and error. Hasil pelatihan menunjukkan konfigurasi terbaik diperoleh pada learning rate sebesar 0,2 dan jumlah neuron hidden layer sebanyak 20 neuron dengan nilai Mean Squared Error (MSE) sebesar 0,0000116. Pengujian model dilakukan secara terbatas menggunakan data hari ketiga dan menghasilkan nilai Mean Absolute Percentage Error (MAPE) sebesar 2,6158%. Nilai error terkecil diperoleh sebesar 0,00334% pada pukul 12.00, sedangkan error terbesar mencapai 24,322% pada pukul 09.15. Hasil tersebut menunjukkan bahwa metode Backpropagation Feed Forward Neural Network memiliki potensi untuk digunakan dalam prediksi jangka pendek energi keluaran PLTS off-grid pada data pengujian yang digunakan. Namun, karena data penelitian hanya mencakup tiga hari pengamatan dan validasi dilakukan pada satu hari pengujian, hasil yang diperoleh masih bersifat studi awal (pilot study) dan belum dapat digeneralisasi untuk berbagai kondisi cuaca maupun musim. Penelitian lanjutan dengan data yang lebih panjang dan beragam diperlukan untuk mengevaluasi keandalan model secara lebih komprehensif.
Off-grid Photovoltaic (PV) systems face challenges in maintaining energy availability due to fluctuations in power generation caused by changing weather conditions. Therefore, an energy forecasting method is required to estimate energy availability for future periods. This study aims to develop an energy output forecasting model for an off-grid PV system using the Backpropagation Feed Forward Neural Network method. The dataset consisted of PV energy output, ambient temperature, and solar irradiance collected from the off-grid PV system at IT-PLN over three consecutive days with a 15-minute sampling interval between 09:00 and 15:00. The neural network architecture comprised 75 input neurons representing energy, temperature, and irradiance data, one hidden layer, and 25 output neurons. The number of hidden neurons and learning rate were determined through a trial-and-error approach. The training results indicated that the optimal configuration was achieved with a learning rate of 0.2 and 20 hidden neurons, resulting in a Mean Squared Error (MSE) of 0.0000116. Model testing was conducted using data from the third day and produced a Mean Absolute Percentage Error (MAPE) of 2.6158%. The smallest prediction error was 0.00334% at 12:00, while the largest error reached 24.322% at 09:15. These findings indicate that the Backpropagation Feed Forward Neural Network method shows potential for short-term forecasting of off-grid PV energy output under the testing conditions used in this study. However, since the dataset covered only three days of observation and validation was performed on a single testing day, the results should be considered a preliminary pilot study and cannot yet be generalized across different weather conditions or seasonal variations. Further studies employing longer and more diverse datasets are required to evaluate the robustness and generalizability of the proposed model.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Muhammad Ilham Amba, Tony Koerniawan, Andi Dyah Harum Hardyanti

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with Journal of Authentic Research agree to the following terms:
- For all articles published in Journal of Authentic Research, copyright is retained by the authors. Authors give permission to the publisher to announce the work with conditions. When the manuscript is accepted for publication, the authors agrees to implement a non-exclusive transfer of publishing rights to the journals.
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution-ShareAlike 4.0 International License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.