<p>This work presents a dual distance-minimizing data-driven algorithm for simulating composite structures comprising both solid and thin-walled components. In regions modeled with solid finite elements, stress–strain data are used to drive the simulation. For thin-walled components discretized with beam, plate, or shell elements, generalized stress–strain data (namely, normal force, bending moment, normal strain, and curvature) are employed to analyze the structural behavior. To connect non-matching meshes arising from different element types and discretization requirements, a penalty-based coupling technique is adopted to enforce displacement continuity and transfer interaction forces across interfaces between solid and structural elements. In this manner, the computational efficiency in thin-walled regions is enhanced by leveraging structural theories, while the accuracy in solid regions is preserved to capture localized stress fields. Several numerical examples are provided to validate the proposed method and demonstrate its robustness for coupling solid and structural elements within the data-driven computing paradigm.</p>