Year: 2025
Type: conference-paper
Abstract: The Beetle Antennae Search (BAS) algorithm demonstrates notable efficiency in heuristic optimization through individual-based search mechanisms, circumventing complex multi-particle computations inherent in swarm algorithms. However, its performance is constrained by unidimensional random search patterns and degraded optimization capacity in high-dimensional spaces. This study proposes a Spatial Dimension-Enhanced BAS (SD-BAS) algorithm that extends random search capabilities into multidimensional spatial domains. For N-dimensional parameter optimization, SD-BAS generates N orthogonal search directions through spatial decomposition, significantly expanding directional diversity. A dynamic step-size adaptation mechanism replaces conventional fixed-step implementations to enhance convergence efficiency. Experimental validation confirms SD-BAS's superior optimization accuracy and convergence speed compared with conventional intelligent algorithms. Practical implementation in PID controller tuning demonstrates significant improvements in parameter calibration efficiency (21.7% faster convergence) and system stability metrics (38.4% overshoot reduction), particularly in high-dimensional optimization scenarios. These advancements position SD-BAS as a robust solution for complex engineering optimization challenges.
Authors: Teng Feng, Yao Mao
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4411887596