Q-REO: Pengoptimal Metaheuristik Hibrida yang Menyeimbangkan Eksplorasi dan Eksploitasi

Alirridlo, Maulana (2026) Q-REO: Pengoptimal Metaheuristik Hibrida yang Menyeimbangkan Eksplorasi dan Eksploitasi. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6025241006-Master_Thesis.pdf Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Masalah optimasi semakin kompleks dan menuntut pendekatan metaheuristic yang mampu secara adaptif menyeimbangkan eksplorasi dan eksploitasi pada berbagai lanskap dan dimensi. Metode yang ada umumnya masih bersifat spesifik, di mana RRT-based optimizer (RRTO) memiliki kemampuan eksplorasi yang kuat namun performanya menurun pada masalah berdimensi tinggi, sedangkan egret swarm optimization algorithm (ESOA) mampu menjaga keseimbangan eksplorasi–eksploitasi dengan baik tetapi memiliki biaya komputasi yang relatif tinggi. Penelitian ini mengusulkan Q-REO, sebuah kerangka kerja hibrida yang mengintegrasikan quasi-reflective initialization, RRTO, ESOA, dan enzyme action optimizer (EAO) untuk menggabungkan keunggulan komplementer masing-masing algoritma ke dalam satu mekanisme adaptif. Q-REO dievaluasi menggunakan 23 fungsi benchmark standar, CEC2017, serta lima masalah desain teknik klasik. Hasil eksperimen menunjukkan bahwa Q-REO memberikan kinerja terbaik pada 111 dari 115 metrik evaluasi dan meningkatkan solusi terbaik dibandingkan ESOA (36 kasus), RRTO (26 kasus), dan EAO (97 kasus). Pada pengujian masalah teknik, Q-REO berhasil memperoleh berat minimum speed reducer sebesar 2994,48 kg dan biaya minimum pressure vessel sebesar 6059,71 USD. Berdasarkan uji peringkat Friedman pada perbandingan hasil dengan dimensi default, Q-REO menempati peringkat pertama dibandingkan seluruh algoritma pembanding, yang menunjukkan keunggulan kinerja dan konsistensi solusi pada skenario evaluasi utama.
=======================================================================================================================================
Optimization problems are becoming increasingly complex and require a metaheuristic approach that can adaptively balance exploration and exploitation across various landscapes and dimensions. Existing methods are generally specific in nature, where RRT-based optimizers (RRTO) have strong exploration capabilities but their performance declines in high-dimensional problems, while egret swarm optimization algorithms (ESOA) are able to maintain a good balance between exploration and exploitation but have relatively high computational costs. This study proposes Q-REO, a hybrid framework that integrates quasi-reflective initialization, RRTO, ESOA, and enzyme action optimizer (EAO) to combine the complementary strengths of each algorithm into a single adaptive mechanism. Q-REO was evaluated using 23 standard benchmark functions, CEC2017, and five classic engineering design problems. Experimental results show that Q-REO delivers the best performance on 111 out of 115 evaluation metrics and improves the best solution compared to ESOA (36 cases), RRTO (26 cases), and EAO (97 cases). In engineering problem testing, Q-REO successfully obtained a minimum speed reducer weight of 2994.48 kg and a minimum pressure vessel cost of 6059.71 USD. Based on the Friedman ranking test comparing results with default dimensions, Q-REO ranked first compared to all comparison algorithms, demonstrating superior performance and solution consistency in the main evaluation scenario.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Algoritma Metaheuristik Hibrida, Algoritma Optimasi, Evaluasi Fungsi Benchmark, Optimasi Berdimensi Tinggi, Optimasi Rekayasa, Benchmark Function Evaluation, Engineering Optimization, High Dimensional Optimization, Hybrid Metaheuristic Algorithms, Optimization Algorithm
Subjects:	Q Science > QA Mathematics > QA9.58 Algorithms T Technology > T Technology (General) > T57.84 Heuristic algorithms.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55101-(S2) Master Thesis
Depositing User:	Maulana Alirridlo
Date Deposited:	02 Feb 2026 07:26
Last Modified:	02 Feb 2026 07:26
URI:	http://repository.its.ac.id/id/eprint/131603

Actions (login required)

View Item