Feasibility- and connected-domain-guided multi-objective particle swarm optimization for multi-loop mechanism synthesis

Abstract Dimensional synthesis of multi-loop mechanisms must satisfy loop closure, branch continuity, singularity margin, task coverage and load transfer requirements. Penalty-based treatment is inefficient for fragmented feasible input domains under limited evaluations. This study formulates constrained sizing as a two-objective search that minimizes task-coverage error and a clipped load proxy, while using connected-domain quality as internal guidance. A feasibility- and connected-domain-guided multi-objective particle swarm optimizer (FCD-MOPSO) is proposed, combining mechanism-aware input-space evaluation, feasibility-first comparison, connected-domain-guided archive maintenance, mixed leader selection and bounded perturbation. Under a soft-constrained low-budget protocol, FCD-MOPSO increased the valid-run rate from 0.6 to 0.8 and the high-quality (HQ) and force-aware HQ rates from 0.1 to 0.4 relative to Basic MOPSO. Search histories showed earlier archive formation and more consistent coverage accumulation. Representative inverse-dynamics verification showed lower peak force, RMS force, force-gradient risk and smoothness index than selected Basic MOPSO and PSO representatives. Strict-feasibility diagnosis identified coverage pressure as the limiting factor. FCD-MOPSO is positioned as a mechanism-aware screening optimizer for fragmented feasible active-input domains.