Automated Synthesis of Planar Linkage Mechanisms with Diverse Joint Types via Spring-Connected Link Models and Contrastive Graph Learning

The automated synthesis of planar linkage mechanisms has long been a challenge in mechanism design, requiring both geometric feasibility and motion accuracy. Recent advances in data-driven and neural network–based methods have shown promise in automating linkage synthesis, improving efficiency and scalability compared to traditional analytical or optimization-based techniques. Nevertheless, existing data-driven approaches remain limited in handling diverse joint configurations and often suffer from unstable optimization when refining linkage geometries. This study proposes a unified framework that combines deep learning with physics-based modeling for the automated synthesis of one-degree-of-freedom (DOF) planar linkages. The method integrates (1) large-scale dataset generation using a spring-connected link model (SLM) capable of representing revolute, prismatic, and pin–slot joints; (2) contrastive graph learning for efficient retrieval of candidate linkages through embedding similarity between paths and graphs; and (3) a two-stage optimization process that refines linkage geometry via global–local search. The SLM enables feasibility evaluation based on strain energy, providing smooth design sensitivities and enhanced optimization stability. Experimental results demonstrate that the proposed method achieves superior accuracy across both seen and unseen target paths, outperforming conventional topology optimization approaches. Overall, the framework offers a promising direction for automating and accelerating the design of complex linkage systems by combining deep learning with advanced optimization techniques.