For indoor renovation robots, obtaining pose information within the architectural blueprint is critical to ensuring the successful execution of renovation tasks. In the construction industry, Building Information Modeling (BIM) serves as a digital tool that provides comprehensive environmental data, including structural details, spatial layout, and renovation requirements. This paper converts BIM into a point cloud map and generates a sequence of simulated sensor data for localization. The localization process is divided into two stages: initial pose estimation using point cloud registration, followed by real-time pose estimation based on a pose graph. Point cloud registration allows the robot to achieve accurate initial localization within the blueprint coordinate system, while pose graph optimization further improves real-time pose accuracy. This method enables direct acquisition of pose information in the coordinate system of the architectural drawings. Experiments conducted in three different scenarios demonstrate that the proposed approach maintains high localization accuracy even under changes in the renovation environment, verifying its feasibility and robustness in practical applications.