With the increasing demand for the use of industrial robots in the industrial field, the design requirements of industrial robots represented by robot arms are also deepening. The MBSE method is applied to the design process of robot arm products. Aiming at the problem of semantic heterogeneity and collaborative disconnection between MBSE and computer-aided design (CAD) models, this paper proposes a robot arm CAD-X Language model transformation method based on X Language metamodel extension. By constructing an X Language metamodel that integrates the geometric topology, physical attributes and process constraints of the robot arm, and establishing the semantic mapping rules between the X Language metamodel and the CAD metamodel, the X Language metamodel is constructed, the automatic transformation from robot arm CAD model to X Language system model is realized. The method includes three stages: 1) parsing geometric, assembly and kinematic features of CAD models based on STEP AP214 and URDF specifications; 2) mapping CAD model elements to X Language extended metamodels through feature recognition and semantic matching; 3) reconstructing X Language model files to support multi-level parameter association and cross-domain semantic traceability. The case study shows that the proposed method can effectively improve the consistency between the system model and the geometric model, reduce the workload of manual intervention, and enhance the traceability of design parameters, which provides technical support for the intelligent collaborative design of complex mechanical and electrical products.