Improved Northern Goshawk Optimization Algorithm for Global Optimization

Global optimization has been used in many real-world problems. Nature-inspired meta-heuristic algorithms, such as the Northern Goshawk Optimization NGO algorithm that has just been proposed, are often used to solve these kinds of optimization problems. An NGO provides satisfactory results. In this algorithm, the proposed exploration model may not provide sufficient coverage of the problem space, trapping the system in a local optimal solution. To improve the performance of NGO, a novel and efficient improved northern goshawk optimization technique named INGO is proposed in this paper. In INGO, a new concept of switching between exploration and exploitation has been developed to improve overall algorithm performance to avoid being stuck in local optima. Also, to increase search capabilities, Levy Flight is used. Twenty-three known benchmark functions were used to test the performance of the proposed INGO. The results were compared to those of an NGO and some well-known robust algorithms. Experimental data indicates that the INGO suggested in this study consistently outperforms the traditional NGO and alternative methods in a significant number of test functions.

I. Introduction

Current real-world optimization tasks are challenging due to their complexity and the enormous dimension of their solution spaces. Heuristic optimization algorithms have been employed in a variety of fields, including engineering, machine learning, scheduling, intrusion detection systems, and formula estimation [1] [2]. In optimization, the goal is to find the best feasible solution from among many different possibilities for a given problem. A multidimensional search problem is typically the result of an optimization procedure. In practice, optimization aims to decrease or increase a fitness function that measures the quality of a solution candidate, which is often represented by a vector in the search area. Meta-heuristics are a class of approximation optimization algorithms that produce reasonable solutions in a timely manner [3]. They are used in science and engineering to tackle difficult and complicated tasks [4].