The Numerical Simulation Of Stress Field In Die-casting Dies And The Application Of BP Neural Network In Optimization Design To Control Deformation Of The Die

With the increase of demand for the diecastings, the defonnation of the die-casting die affected the quality of castings seriously. Aiming at these complexion, this paper studied the influence of structural parameters and processing parameters on the deformation of the dies. By means of the numerical simulation we simulated the deformation of die when the structural parameters and processing parameters are pre-decided, and protested the deformation situation of the die by the results of simulation. So, this project plays is significant importance to improve the quality of casting and reduce the cost of the castings practically.The work mainly took the commercial software ANSYS as simulation platform, The deformation of die to produce the casting which specification was ZL101 alloy, and the cavity projected area was the 300cm~2 in die-casting processing was simulated. The simulation results described the relationship between the defonnation of the die and the flashes. Matlab software was taken as the second platform, which combined the numerical simulation result, simulated and trained the sample data . We can forecast the max deformation of the insert moving part of the die, and achieved to the goal of optimization design of the die-casting by utilizing the calculated results of the BP neural network. The research shows that simulation and analysis of stress field with the technology of numerical simulation can forecast the deformation of die. We found that the injection pressure, distance of cushion block, thickness of support plate seriously influence the deformation of die; but the thickness of cushion block slight influence the deformation of die. By changing the the injection pressure, distance of cushion block, thickness of support plate, the deformation of die can be controlled. Using the result of the numerical simulation in BP neural network during design of die can develop the efficiency of the numerical simulation enormously. Based on neural network training, the result of simulation under other parameters can be forecasted and the relationship of the result and structural parameters can be expressed by charts. As the result, the designer can design die refering to the results of this paper. It not only shortens the design period but also reduces the designer's working load. So, this subject is actual valuable.The results of this project shows that during the design of die, the combination of the numerical simulation and an neural network can make us to grasp the inner rule of various factors with the least experimentation, to obtain accurate result of prediction and optimization, and to achieve the aim of optimization. At the moment, few references reported that the technology of neural network was used in the design of die-casting dies. Therefore this paper possesses the characteristics innovation and the applicable value in some extent.