To address the limitations of traditional manual inspection in construction quality control of deep foundation pits, including poor timeliness, limited visualization, and excessive reliance on inspectors’ experience, a construction quality inspection method integrating augmented reality (AR), building information modeling (BIM), and knowledge base technologies is proposed. Image-assisted localization and point cloud registration techniques are employed to achieve high-precision registration between the BIM model and the construction site in an AR environment, thereby enabling accurate alignment of the virtual model with the real-world scene. A semantically enriched BIM model for deep excavations is established and used as the information carrier to construct a BIM-integrated knowledge base for quality inspection. An intelligent virtual-real augmented construction quality inspection system is developed, providing inspection assistance through virtual-real scene integration, multi-source knowledge retrieval, and visual feedback. Validation through an engineering case study demonstrates that the proposed method is effective and practical, providing a reliable technical approach for intelligent and refined quality management of deep excavation construction.