A Merging-Assembly Strategy for Complex Scenario-Oriented Digital Twin Modeling

Addressing the pervasive issues of structural redundancy and scalability in complex digital twin environments, this paper introduces a novel merging-assembly framework. Conventional digital twin construction typically depends on isolated, black-box integrations, which inevitably lead to severe static resource duplication, high computational costs, and inflexible system architectures. To overcome these barriers, an object-oriented modeling-driven approach (OOM) is developed in this paper, grounded in the five-dimensional digital twin architecture (5D-DT). The merging phase targets identical and similar sub-models, systematically extracting shared prototypes at the physical property modeling layer to eradicate geometric and logical repetitions. Following this, the assembly phase dynamically reconstructs the functional system topology via unit-level fusion and line-level connection, ensuring robust synchronization under varying spatiotemporal constraints. Validation within a pharmaceutical packaging context confirms the framework's industrial viability. By leveraging shared the physical property modeling layer configurations and optimized connection structures, the system effectively controls redundant resource consumption. Specifically, the proposed method achieved an approximately 27% reduction in Video RAM (VRAM) occupancy, accompanied by decreases in physical storage requirements and system initialization latency. Ultimately, this methodology offers a viable exploration for developing scalable and lightweight complex cyber-physical systems (CPS), aiming to improve asset reuse without compromising functional fidelity.