OptimumDesignofCombinedColdExtrusionDieforBevelGear.pdfVIP

IACSIT International Journal of Engineering and Technology, Vol. 4, No. 4, August 2012 Optimum Design of Combined Cold Extrusion Die for Bevel Gear Qing-hua Yang, Xin Chen, Bin Meng, and Juan Pan ? Abstract—A collaborative optimization method for the structure parameters of combined die for bevel gear cold extrusion forming is presented in this paper, combined using finite element analysis, orthogonal experiment, neural network and genetic algorithm. Orthogonal experiment is used to design experimental schemes. Neural network is used to establish mapping relationship between die and process parameters and maximum extrusion force. Genetic algorithm is used to optimize the structure parameters.FEA evaluation validates that the collaborative optimization method presented has the same accuracy as that of conventional FEA based optimization approach, while it overcomes the deficiency of large computational resource consumption since the finite element analysis is relatively independent of optimization process and just provides training samples of neural network and evaluates the optimized results obtained. The collaborative optimization method can provide a fast and effective approach for combined die. Index Terms—Combined die, optimum design, neural network, genetic algorithm. I. INTRODUCTION The pressure on die can be high enough to reduce the strength and life of die in the process of cold extrusion forming, thus correct choice of the materials and heat treatment methods, correct design of structure and optimization parameters for die are important, and the latter is of great significance[1]. Artificial intelligence algorithms such as neural network and genetic algorithm are widely used in the optimization of cold extrusion forming such as the structure parameters of die and the process parameters. The neural network has capabilities of self-learning, self-organization, self-adaptation and non-linear dynamic processing [2~4]. The training samples need to be input and