Identifying Solid-Fluid Parameters of the Subsea Tunnel Based on Evolutionary ANN

How to determine the fluid-solid geophysical parameters of subsea tunnel is the basement of calculation and construction. Aiming at the problems of long time of forward numerical calculation and locally optimal solution, the paper proposed an identification method for fluid-solid parameters based on difference evolution (DE) arithmetic and artificial neural network (ANN). This paper adopts orthogonal experimental design and numerical simulation to produce learning samples, utilizes the nonlinear reflection of ANN and global optimization of DE, then establishes the evolution ANN model relating the solid-fluid parameters and character indices. Furthermore, we use the relative error between predictive and observed character indices as fitness of DE, searches the solid-fluid parameters in agreement with observation. This method can utilize the large engineering software. The calculation case study states that the method is feasible and can get satisfactory result.