Risk assessment method of water inrush in tunnels constructed in water-rich grounds

A risk assessment method of water inrush in tunnels constructed in water-rich grounds, comprising the following steps:

• step 1:
• first, simulating a tunnel excavation process by finite element software and fluid-structure interaction, building a three-dimensional model of tunnel construction according to field construction data, then analyzing precursory information and stress-displacement response of a water inrush accident occurring in subway tunnel construction according to the three-dimensional model of tunnel construction, and determining causes of the water inrush accident occurring in a fracture zone of a water-rich ground according to analysis results;
• step 2:
• according to a research method of control variables, creating a tunnel construction model under different working conditions by the finite element software, qualitatively and quantitatively analyzing effects of factors of a groundwater level, an elastic modulus of a fracture zone and advanced pipe shed grouting on the stability of surrounding rock, and summarizing change rules of deformation and seepage of the surrounding rock under different working conditions;
• and
• step 3:
• based on the field construction data, acquiring physical and mechanical indexes of soil and rock of a complex stratum in an offshore area and engineering characteristics under a condition of rich water, summarizing water yield properties of various kinds of complex strata, improving an algorithm of a radial basis function (RBF) neural network using a Grey Relation Analysis (GRA)-based Partitioning Around Medoid (PAM) clustering algorithm to assess risks of collapse and water inrush occurring in the stratum, building a water yield property assessment model according to the causes of the water inrush accident and the change rules of deformation and seepage of the surrounding rock under different working conditions, extracting a training sample and a test sample from actual tunnel engineering, training the built neural network model, and assessing the risks of water inrush in tunnel construction.