To enable city-wide, unified management and deep integration of engineering project data, this study addresses the technical bottlenecks in merging multi-source, heterogeneous BIM and CIM datasets. A unified spatiotemporal datum coupled with a high-precision spatial index is established to align cross-scale data and ensure accurate positioning. A lightweight-yet-fidelity-preserving optimization strategy—based on geometry–attribute separation and multi-level LOD—is introduced to significantly boost data processing and rendering performance without sacrificing visual realism. A semantic-mapping and dynamic-update mechanism is then designed to create an uninterrupted, life-cycle data chain spanning planning, design, construction, acceptance, operation and maintenance, and urban renewal. The Wuhan BIM Management Information Platform demonstrates how the proposed fusion technologies perform in real-world business scenarios and effectively support inter-departmental, cross-stage collaboration. Results confirm the feasibility of the presented technical system and offer a replicable methodology and practical paradigm for building digital-twin cities.