摘要
建立了改进的BP神经网络、数值仿真和遗传算法相结合的铝型材挤压模工作带长度优化模型。将型材截面划分单元 ,由正交试验法得到单元工作带长度值作为网络训练样本的输入值 ,模型目标值为变形后质点速度均方差。采用基于有限体积法的数值仿真技术获得样本目标值。模型的全局优化解由遗传算法求得。
An optimization model for designing the die bear in g of aluminum extrusion is presented, which integrates ameliorated BP neural net work,numerical simulation and genetic algorithm. The area of extrusion section is divided into several units and the bearing values of them are given as the in puts of network training specimen by using the orthogonal method. The target val ue of the model is velocity mean-squared error after forming. Finite volume met hod is used in the numerical simulation to get the target value of specimen and the general optimized solution is attained through genetic algorithm.
出处
《机械科学与技术》
CSCD
北大核心
2004年第9期1015-1018,共4页
Mechanical Science and Technology for Aerospace Engineering
关键词
BP神经网络
遗传算法
有限体积法
铝型材工作带优化
BP neural network
Genetic algorithm
Finite volu me method
Optimization of aluminum die bearing
