Tribology Optimum Design Of Extrusion Die

The design of extrusion-die is a very complicated process which needs compositive analysis and repeated operations. The design with modeling, optimization and flexibility is developing toward integration, intelligence automatization. Die life is an all-around technology problem. And wear is the predominant factor which affects the die life, especially die life at a high temperature is affected by wear in over 70%. So, in hot extrusion process, how to reduce wear and improve die life is one of the most concerned problems for mold designer.This paper put forward design optimization of extrusion-die system, which integrated existing engineering numerical analysis, BP Neural Network and genetic algorithms. First, based on updated Archard's theory, the paper calculated wear depth by FEM software. Then, combining Neural Networks with FEM, the paper made data of FEM simulation in light of learning sample and trained model of BP Neural Networks, which can forecast mold wear and rapidly forecast life of mold. Resource of knowledge of optimum design is gained after training end and simulation test validate reliability that the model reflected practical problems. During Genetic Algorithm iterative process, the fitness values are obtained on the basis of multilayer BP Neural Network and its popularized capability. It resolves limitation that the time of simulation is too long and the calculation- quantity of wear is too mass. Meanwhile, this method affords a kind of optimum design of extrusion-die. Because the numerical computation technology, Artificial Neural Network and Genetic Algorithm are not coupling each other, it is a kind of feasible method and has much engineering value to be applied.