DSD-LI-PO: A Bioinspired Parrot Optimizer with Componentwise Swarm Hybridization and Contracted Lens Imaging Refinement for Constrained Engineering Optimization

Bioinspired swarm optimizers are widely used for constrained engineering optimization, yet whole-vector updates can disrupt well-evolved components and insufficient elite refinement can limit late-stage accuracy. This study proposes DSD-LI-PO, an enhanced Parrot Optimizer (PO) integrating dynamic swarm dimension hybridization (DSD) with stagnation-triggered variable precision lens imaging (LI). DSD selectively inherits components from the current best solution to preserve useful dimensional information, whereas LI generates a progressively contracted elite-derived candidate when stagnation occurs. DSD-LI-PO is evaluated on CEC2017, six representative functions at 30, 50, and 100 dimensions, ablation variants, and five constrained engineering design problems. On CEC2017, it ranks second overall with an average rank of 1.8966 and records 28/1/0 wins/ties/losses against PO and 29/0/0 against IPO. In the scalability tests, it remains second in all three dimensions, with average ranks of 1.9167, 1.9167, and 1.7500. The complete method ranks first in the ablation study and second in the engineering tests, with feasible solutions in all runs. These results demonstrate that DSD-LI-PO provides a consistent and targeted improvement over PO while maintaining competitive performance against strong non-PO optimizers.