Wind Farm Layout Optimization Problem (WFLOP) aims to optimize turbine placement to mitigate wake-effect losses and improve power generation efficiency. Existing particle swarm optimization (PSO)-based approaches for WFLOP mainly rely on conventional variants and lack problem-specific search mechanisms. To address this limitation, this paper proposes a level-direction guided hierarchical particle swarm optimization algorithm (HGPSO). The proposed method integrates multi-direction guidance and hierarchical learning to enhance population diversity and convergence performance. Comprehensive experiments under various wind conditions, turbine scales, and land constraints show that HGPSO outperforms several state-of-the-art WFLOP optimizers in most test cases, achieving average improvements of approximately 0.5%–3.0% in energy conversion efficiency. Statistical tests further confirm its robustness and stability. The proposed HGPSO provides an effective and scalable solution for complex wind farm layout optimization.