Modeling the dynamic safety management of buildings adjacent to karst shield construction: An improved cloud Bayesian network

Abstract

Shield construction in karst tunnels inevitably causes settlement deformation and even damage to surrounding buildings. To accurately evaluate the security of neighboring buildings, a hybrid method combining an improved cloud model (CM) and a Bayesian network (BN) is proposed for dynamic security analysis and the evaluation of neighboring buildings. First, through engineering practice, simulation analysis and literature research, the safety risk factors and evaluation criteria for adjacent buildings influenced by shield construction in karst areas are determined, and a BN for safety evaluations of adjacent buildings is constructed. The improved CM is used to model the uncertainty of the risk state, realize the discretization of continuous data and obtain prior knowledge. By improving the cloud Bayesian network (CBN) to perform three kinds of comprehensive analyses, including forward prediction reasoning in advance, sensitivity analysis and reverse risk diagnosis for events, and appropriate risk management and control measures, the dynamic safety perception and control of buildings adjacent to karst shield construction is realized. The data for buildings under construction adjacent to Guiyang Rail Transit Line 3 are used to demonstrate the applicability and effectiveness of the proposed method. The contributions of the research results are as follows: (a) The key risk factors affecting the safety of buildings adjacent to shield construction areas in karst zones are extracted, and an evaluation index system and an evaluation standard are established; (b) An improved CM is used to describe the fuzziness and uncertainty in the process of node state discretization in a risk inference network, thus improving the prediction and diagnosis performance of BN security evaluation models; (c) The improved cloud Bayesian model is applied to construct a knowledge representation-based safety model of buildings adjacent to karst shield construction areas, and accurate perception and control diagnosis results are obtained for effective safety risk prediction and control, providing a reference for other similar projects.