Ashraf Al-Ouʼn, Saleh Ali Alomari, Raed Abu Zitar, Mirjalol Ismoilov, Aseel Smerat, Widi Airbow, Zeinab Montazeri, Mohammad Dehghani, Om Parkash Malik, Kei Eguchi
The Numbat Optimization Algorithm (NOA), a metaheuristic inspired by the distinctive foraging ecology of the numbat, an insectivorous mammal exhibiting wide-range exploration, olfactory-guided directional search, and highly precise point-digging behaviour, is introduced. The core novelty of NOA lies in the rigorous and behaviour-consistent mapping of these biological characteristics into a mathematically structured multi-phase optimization framework that achieves an effective balance between exploration and exploitation without requiring complex control parameters. In the NOA computational structure, each member of the ensemble is defined as a candidate option for achieving the optimal response and its position is modified and upgraded in a systematic, continuous, and multi-stage process. This evolutionary mechanism is based on four independent processes, which include (1) comprehensive exploration of the problem domain with the aim of revealing promising parts, (2) dynamic and adaptive assessment of the surrounding conditions and instantaneous correction of the movement direction in case of a decrease in the value of sensory data, (3) targeted and precise penetration into selected areas through deep local improvements, and finally (4) controlled revision of states with a history of favorable performance in order to strengthen areas with higher capacity in the response space. To analyze the internal dynamics of the NOA search process, population diversity analysis, exploration, and exploitation using well-known diversity indices have been performed, which show that NOA maintains high diversity in early iterations and gradually increases exploitation intensity as convergence progresses. Extensive simulation studies have been conducted on twenty-three standard benchmark functions including unimodal, high-dimensional multimodal, and fixed-dimensional multimodal problems. The findings obtained from the simulation results show that NOA ranks first in all unimodal functions, five out of six high-dimensional multimodal functions, and nine out of ten fixed-dimensional multimodal functions, and has an overall superiority in 21 out of 23 benchmark functions. These results confirm the efficiency, robustness, and high optimization capability of the NOA algorithm in dealing with challenging and diverse optimization problems. © This article is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. License details: https://creativecommons.org/licenses/by-sa/4.0/
Department of Computer Science, Faculty of Information Technology, Al al-Bayt University, Mafraq, 25113, Jordan; Computer Science Department, Faculty of Information Technology, Jadara University, Irbid, 21110, Jordan; College of Engineering and Computing, Liwa University, Abu Dhabi, 41009, United Arab Emirates; Department of Transport systems, Urgench State University named after Abu Rayhan Biruni, Urgench, 14, Kh. Alimdjan str, Urgench city, 220100, Uzbekistan; Hourani Center for Applied Scientific Research, Al-Ahliyya Amman University, Amman, 19328, Jordan; Department of Electrical Engineering, Faculty of Vocational Studies, Universitas Negeri Surabaya, Indonesia; Department of Electrical and Electronics Engineering, Shiraz University of Technology, Shiraz, 7155713876, Iran; Department of Electrical and Software Engineering, University of Calgary, Calgary, T2N 1N4, AB, Canada; Department of Information Electronics, Fukuoka Institute of Technology, Japan
