PDPSO: Priority-driven search particle swarm optimization with dynamic candidate solutions management strategy for solving higher-dimensional complex engineering problems

Particle swarm optimization (PSO) is one of the most classical meta-heuristic algorithms (MAs), and the algorithm is extensively used in solving practical engineering application problems. To address the challenges of particle swarm optimization in high-dimensional complex engineering problems, including accuracy, stability, and resource utilization, we propose a PSO variant called PDPSO, which incorporates a priority-driven search strategy and a dynamic candidate solution management strategy. The priority-driven search strategy removes the inertia term, retaining the guidance of the individual optimal and high-priority candidate solutions, significantly enhancing algorithm stability, robust and execution efficiency; the dynamic candidate solution management strategy constructs an adaptive priority table, dynamically screening high-quality solution sets via a non-uniform subpopulation allocation mechanism, combined with a state rollback mechanism to prevent local stagnation, and integrating candidate solution elimination to achieve a self-balancing equilibrium between algorithm exploration and exploitation. We have not added any mutation strategy to the original algorithm, instead, the PDPSO algorithm balances the exploration and exploitation capabilities by whether or not to continue to learn the previous candidate solutions at the next update, which is a significant highlight of the algorithm and provides a new research direction to improve the robustness of other optimization algorithms for high-dimensional complex problems. Verified by CEC2017 high-dimensional testing (30, 50, and 100 dimensions) and 34 engineering examples, the PDPSO algorithm exhibits outstanding advantages in high-dimensional problems and excellent computational efficiency. When the test function dimension increases from 30D to 100D, the growth rate in time complexity is as low as 24.87%. The PDPSO algorithm demonstrates strong applicability in real-world engineering problems, with optimization results in various complex simulation scenarios approaching theoretical optimal solutions. PDPSO provides a high-precision, robust solution for high-dimensional engineering optimization, and its dimension-insensitive complexity characteristics have universal value. The source codes and supplementary materials of PDPSO are available at https://github.com/hepeidong1/PDPSO.