PSOX: Leveraging Historical and Global Optima to Enhance Diversity and Speed in Genetic Algorithm Crossover

Abstract

This study introduces PSOX, a novel crossover operator for real-coded genetic algorithms, strategically integrating particle swarm optimization (PSO) principles to revolutionize search efficiency. Unlike conventional crossover methods limited to intra-generational information exchange, PSOX leverages guidance from both the global best solution and historical optima across generations, enabling adaptive exploration of promising search regions while preserving population diversity. Rigorous evaluations on 15 benchmark functions—spanning unimodal, multimodal, and highly complex landscapes—demonstrate that PSOX outperforms five state-of-the-art operators in solution accuracy, stability, and convergence speed, particularly when paired with optimal mutation strategies. A detailed analysis of mutation rate impacts provides actionable parameter-tuning insights for diverse optimization scenarios. By bridging GA’s evolutionary robustness with PSO’s directional learning, PSOX offers a breakthrough for accelerating convergence in complex problem domains.