The traditional substation operation and maintenance model is facing challenges such as data dispersion, response lag, and low 3D visualization. The 3D reality interaction system becomes the key to break the situation. By analyzing the standard model library, the study generates a point cloud uniformly and randomly in the triangular grid of basic elements, and then assembles it into a complete component model based on the transformation matrix. OPTICS clustering algorithm is used to segment the complex distribution of point clouds of buildings in the scene. And the integration of Web front-end technology, BIM modeling and operation and maintenance strategy optimization, to build a three-dimensional reality interactive system, to achieve equipment status visualization and inspection path planning. Under the harsh condition of point cloud thinning to 50%, the model coverage rate can still maintain 74.20%. The triangulated point cloud + OPTICS clustering in this paper significantly outperforms PCA+KNN and RANSAC+ICP with an average recognition time of 6.07 s and an accuracy of 96.73%, and the recognition accuracy shows stronger anti-interference ability under the interference of added noise and uneven density, with a noise variance of 0.08, which still reaches 55.68%. The average frame rate of 3D model rendering reaches 39.94 fps, running smoothly and stably. In actual operation and maintenance tasks, the system improves the efficiency of operations such as “running to cold standby” by 39% to 52%.
