DECISION SUPPORT ANALYSIS FOR SAFETY ASSURANCE IN METRO CONSTRUCTION BASED ON FUZZY BAYESIAN NETWORKS
This paper presents a novel model to assess the risk of adjacent buildings in tunnelling environments based on Extended Cloud Model (ECM). ECM is an organic integration of Extension Theory (ET) and Cloud Model (CM), where ET is appropriately employed to flexibly expand the variable range from [0, 1] to (-∞, +∞), and CM is used to overcome the randomness and fuzziness during the gradation of evaluation factors. The risk level of a specific adjacent building is assessed by the correlation with the cloud models. A confidence indicator θ is proposed to illustrate the rationality and reliability of evaluating results. Ten buildings adjacent to Wuhan Metro Line Two (WMLT) are randomly chosen among hundreds of adjacent buildings for a case study. Results have proved to be consistent with the actual situation. Compared with other traditional evaluation methods, ECM has been verified to be a more competitive solution with high calculation accuracy, wide adaptability, as well as simplified computer programming. There are no stringent requirements on the quantity of training data during the modelling process, and the original data can be directly entered into ECM without a normalization procedure, avoiding the potential information loss. ECM can be offered as a decision support tool for the risk assessment in urban tunnelling construction and worth popularizing in other similar complicated projects.
Evaluation; Evaluations; Model; Models; risk assessment; Environment; Tunnelling; Risks; Risk; Data;