Predigest Computation Of Welding Finite Element Method Simulation Base On Artificial Neural Network

In modern manufacturing industry, welding has been widely used. Welding finite elements method(FEM) simulation becomes an important study field. Welding process numerical simulation always cannot fit the actual process or simulation time is too long for nonlinear factors such as fields coupling computation and varying parameters. In this article,artificial neural network(ANN) is introduced into FEM to simplify model or connect classes of models. Try to solute some problems of welding simulation.1.Introduce ANN into plates welding process simulation try to acquire relationship of different size welding simulation models with same process parameters. Use this method try to give a simple answer of large welding model simulation.2.ANN method aided FEM in tube-plate welding simulation by divided simulated process into two processes. ANN mapped plates welding residual stress field to the whole tube-plate welding model as the initial stress field. The problem of large number of meshes in simulation and mesh distorted are avoided using this method.