Neural-Network Based Energy Management System for Battery-Ultracapacitor Hybrid Storage

Yusvianti, Fabria Alieftya (2024) Neural-Network Based Energy Management System for Battery-Ultracapacitor Hybrid Storage. Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 6022221004-Master_Thesis.pdf] Text
6022221004-Master_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2026.

Download (12MB) | Request a copy

Abstract

As the need to reduce carbon dioxide (CO2) emissions grows, clean energy solutions such as standalone photovoltaic (PV) systems paired with energy storage offer promising solutions. Batteries are typically used to store surplus energy due to their high-energy density, which can lead to increased stress and reduced lifespan when subjected to sudden changes in irradiation and load. Combining them with ultracapacitor, which have high-power density, can alleviate battery stress and extend their lifespan.

This research explores the integration of a battery-ultracapacitor hybrid energy storage system with standalone PV setups, emphasizing the application of neural network-based management for optimizing power distribution and accelerating output prediction. The study evaluates various scenarios by varying the State of Charge (SOC) of both the battery and ultracapacitor components. Comparisons are drawn between target outputs from reference research optimization and outputs generated by the neural network across 35 test cases. Results indicate that the neural network effectively manages power sharing with a normalized Root Mean Square Error (RMSE) of under 5% and an elapsed time simulation of under one second.

====================================================================================================================================

Dengan meningkatnya kebutuhan untuk mengurangi emisi karbon dioksida (CO2), solusi energi bersih seperti sistem fotovoltaik (PV) mandiri yang dipasangkan dengan sistem penyimpanan energi menawarkan solusi yang menjanjikan. Biasanya, baterai digunakan untuk menyimpan energi berlebih karena memiliki kepadatan energi tinggi, yang dapat menyebabkan stres yang meningkat dan umur pakai yang lebih pendek saat terjadi perubahan mendadak dalam radiasi dan beban. Menggabungkannya dengan ultrakapasitor, yang memiliki kepadatan daya tinggi, dapat mengurangi stres pada baterai dan memperpanjang umur pakainya.

Penelitian ini mengeksplorasi integrasi sistem penyimpanan energi hibrida baterai-ultrakapasitor dengan pengaturan PV mandiri, dengan menekankan penggunaan manajemen berbasis neural network untuk mengoptimalkan distribusi daya dan mempercepat prediksi output. Studi ini mengevaluasi berbagai skenario dengan memvariasikan State of Charge (SOC) dari komponen baterai dan ultrakapasitor. Perbandingan dilakukan antara output target dari optimisasi penelitian referensi dan output yang dihasilkan oleh neural network dalam 35 kasus uji. Hasil penelitian menunjukkan bahwa neural network efektif mengelola pembagian daya dengan nilai kesalahan Root Mean Square Error (RMSE) yang dinormalisasi di bawah 5% dan simulasi waktu yang berlangsung kurang dari satu detik.

Item Type: Thesis (Masters)
Uncontrolled Keywords: battery, neural network, hybrid energy storage, hybrid energy storage system, ultracapacitor, baterai, neural network, penyimpanan energi hibrida, sistem penyimpanan energi hibrida, ultrakapasitor.
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK2941 Storage batteries
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User: Fabria Alieftya Yusvianti
Date Deposited: 28 Jul 2024 13:20
Last Modified: 28 Jul 2024 13:20
URI: http://repository.its.ac.id/id/eprint/109392

Actions (login required)

View Item View Item