Intelligent design method of mine tunnel portal driven by knowledge graph

• Published in: Computers and geotechnics Vol. 173; p. 106431
• Main Authors: Wu, Jiaming, Xiao, Mingqing, Dai, Linfabao, Bo, Huajun, Lian, Zhixiang, Zhou, Hao, Yang, Jian, Pu, Jianwei, Cheng, Hongzhan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2024
• Subjects: BIM; Graph neural networks; Intelligent design; Mine tunnel portal; Tunnel portal knowledge graph; Mine tunnel portal; Graph neural networks; Intelligent design; Tunnel portal knowledge graph; BIM
• ISSN: 0266-352X; 1873-7633
• DOI: 10.1016/j.compgeo.2024.106431

The mine tunnel portal holds significant importance intunnel engineering. However, the tunnel portal design is governed by many factors, often relying on manual design experience. In response to the challenge that the portal design based on manual experience lacks technical transitivity, an intelligent design method of tunnel portal driven by knowledge graph and graph neural networks is proposed. Firstly, through the analysis of portal design influence factors, a comprehensive collection of design cases is undertaken, enabling the establishment of a knowledge graph. Considering the interaction between the entities in the knowledge graph, the characteristics are extracted through Multilayer Perceptron (MLP) and Gate Recurrent Unit (GRU). Considering design experience of the knowledge graph, the convolution layer is implemented to capture the attributes. Fully Connected Network (FC) is utilized for fusion and dimensionality reduction, and the prediction regarding portal type and number of opening holes is effectively achieved. The effectiveness and superiority of this method are analyzed through several sets of experiments. Furthermore, the design parameters predicted are integrated into three-dimensional representation utilizing BIM technology. This model realizes the innovation in the design method for tunnel portals, resulting in enhancements in design efficiency while promoting the advancement of tunnel intelligence.