Construction and Application of Three-dimensional Information Management System for Intelligent Buildings Integrating BIM and GIS Technologies

Main Article Content

Jing Shi

Abstract

Smart building technologies are widely used in all aspects of building structure, services, and management, helping to create a more comfortable, safe and convenient building environment. Building Information Modelling (BIM) and Geographic Information System (GIS) technologies are both widely used intelligent building technologies, and their combination can improve the analytical ability of spatial environment to a certain extent. However, it is difficult to manage them due to the huge amount of data in the Three-Dimensional (3D) information of intelligent buildings. Therefore, it is very important to improve the information management ability of intelligent building 3D information management systems (Moballeghi et al. 2023; Mahamood and Fathi 2022). BIM and GIS technologies were used to build a 3D information management system for intelligent buildings more effectively. The design and development principles of the information management system were explained, and the overall framework of the system was also designed. Research was conducted on feature extraction and matching through an improved scale invariant feature transformation algorithm to enhance the information classification and management capability of the intelligent building 3D information management system. In addition, the improvement measure for SIFI  algorithm was to reduce pixel processing to reduce its memory size. The study explained the preprocessing of model normalization before feature extraction and matching. The coordinate system rotation normalization of building 3D models was achieved through principal component analysis. Finally, the calculation of covariance matrix was explained. The number of pyramid image groups was adopted to further improve the scale space and enhance computational efficiency. The Hessian matrix was introduced to eliminate unstable fixed points. And the purity of feature point matching through similarity coefficients was improved. In addition, a modified multi-view  convolutional neural network was used to classify the feature data, and a modified classification architecture was designed to build a 3D model based on this algorithm to enhance its information classification management capabilities. The study explained the calculation of view weights and global descriptors and described the fully connected and classification architectures. The results showed that the improved scale-invariant feature conversion algorithm achieved a matching accuracy of 98.3% and takes only 17 s. Meanwhile, the proposed multi-view convolutional neural network achieved an accuracy of 97.6% and an F1 value of 96.4% for the classification of 3D information of intelligent buildings. Among the six types of 3D building models selected, the method achieved the highest accuracy of 94.26% and was more stable. It shows that the proposed method of 3D information management of intelligent buildings has obvious classification advantages and provides a new technical reference for the information development of intelligent buildings.

Article Details

Section
Special Issue - Data-Driven Optimization Algorithms for Sustainable and Smart City