For high-dimensional binary optimization problems, classical Gravitational Search Algorithm (GSA) often suffers from premature convergence, stagnation, and diversity loss due to its global gravitational constant and shrinking \(K_{best}\) neighborhood structure. To address these limitations, this paper proposes a Neighborhood Archive-Based Binary Gravitational Search Algorithm (NBGSA) that integrates a self-adaptive gravitational scaling mechanism with two complementary neighborhood archives: a distance-based D-archive and a fitness-hierarchical F-archive. The D-archive preserves local diversity by guiding candidate interactions based on spatial proximity, while the F-archive enforces structured directional convergence through fitness-level hierarchy. Candidate-specific gravitational scaling modulates the influence of neighbors according to both fitness and relative distance, ensuring stable yet intensified exploration of high-quality regions. The effectiveness of NBGSA is demonstrated on a 100-dimensional multi-objective wind-farm layout optimization problem in a binary search space, using two real-world wind datasets. Numerical results demonstrate that NBGSA achieves superior turbine placement and improved optimization performance, particularly for layouts with more than 30 turbines, outperforming recent binary GSA variants and MOEA/D.
