A novel metaheuristic algorithm using structured population and virtual particles

Erik Cuevas

doi:10.1007/s11227-025-07194-x

A novel metaheuristic algorithm using structured population and virtual particles

2025-04-22

0

PRE

AI

E

Erik Cuevas *

Ó

Óscar A. González-Sánchez

N

Noé Delgado-Castañeda

D

Daniel Zaldívar

A

Alma Rodríguez-Vázquez

DOI：10.1007/s11227-025-07194-x

Original

Original request for help

Save

Abstract

En 中文

Conventional metaheuristic algorithms frequently rely on unstructured and fixed populations, neglecting the solution memory and treating candidate solutions in their traditional form. These limitations can result in inefficiencies in the search process, such as revisiting low-quality regions and reducing the exploration flexibility. This paper introduces the Structured population algorithm (SPA) as a novel metaheuristic algorithm that addresses these challenges by integrating a structured population, a solution list, and virtual particles to enhance the search process. SPA commences by initializing a set of virtual particles that represent positions in the search space but do not directly correspond to solutions. Each virtual particle generates a subset of solutions within a defined neighborhood and these solutions are grouped according to their originating particles. The algorithm operates iteratively by selecting the top three solutions from each group to form a local triangle and calculating the pivot points as the centroids of these triangles. Subsequently, the top three pivot points of all local triangles were identified to construct a global triangle. Using this information, a global centroid is derived to guide the movement of virtual particles, with the worst-performing particle moving in the opposite direction to enhance exploration. Updated particles generate new solutions in their neighborhoods, iteratively refining the population until convergence. The performance of the algorithm was validated using shifted test functions designed to assess center-biased tendencies, demonstrating competitive results with superior solution quality, faster convergence, and effective handling of the center-biased limitations.

Keywords:

Metaheuristic algorithms

Structured population

Virtual particles

Optimization

Simplex

Center-biased operators

A novel metaheuristic algorithm using structured population and virtual particles

Abstract

Journal

Organization

Cited Papers