Hybridizing Particle Swarm Optimization with the Filled Function Method for Enhanced Global Optimization Performance

Abstract

This research proposes a novel hybrid optimization algorithm that integrates Particle Swarm Optimization (PSO) with the Filled Function Method (FFM) to address the persistent challenges of premature convergence and local optima entrapment in high-dimensional, multimodal optimization problems. The hybrid PSO–FFM algorithm incorporates a one parameter filled function that dynamically modifies the search landscape when stagnation is detected, enabling systematic escape from deceptive basins while maintaining PSO’s exploration-exploitation balance. Experimental evaluation across eight benchmark functions (Sphere, Rosenbrock, Rastrigin, Griewank, Ackley, Schwefel, Zakharov, and Alpine1) demonstrates significant performance improvements, with the hybrid algorithm achieving up to 100% enhancement over standard PSO in several cases. The proposed method offers a robust framework for complex optimization tasks, particularly in engineering and computational applications where traditional metaheuristics struggle with complex search landscapes.