Kinematic Analysis of Structural Mechanics based on AI Agent

Kinematic analysis is a foundational task in structural mechanics for determining the geometric composition and stability of planar bar systems. This paper develops an agent system based on the LangChain framework for structural mechanics kinematic analysis. The system takes a large language model as its reasoning core and integrates two skills: one is the doubao-visualization skill package, which provides image understanding and rendering capabilities to recognize input structural example images and draw schematic diagrams annotated with analysis results; the other is the structural-kinematic-analysis professional skill package, built according to structural mechanics textbooks, which can parse nodes, bars, and supports, compute the degrees of freedom of the system, and invoke rules such as the binary-body, two-rigid-plate, and three-rigid-plate rules to perform kinematic analysis. By merely inputting a structural image, the agent can autonomously plan and invoke the required skills, output a verdict on whether the system is geometrically stable or unstable together with visualized results, and, thanks to its memory capability, support interactive follow-up questions such as "Is it still stable if a certain support is removed?" Tests on two typical examples of a geometrically stable system and a geometrically unstable system show that the agent produces correct results, verifying the feasibility of the proposed approach and offering a new pathway toward intelligent assistance for structural mechanics teaching and the preliminary assessment of engineering structures.