基于混合神经网络与UG/KF方法的浇口位置优化设计

在分析遗传算法特点的基础上,针对遗传算法优化搜索过程中收敛速度慢和不成熟收敛的缺点,提出一种改进的实数编码混合遗传算法,借助于知识熔接技术即KF语言进行几何建模,建立基于混合神经网络与UG/KF方法的浇口位置优化设计,利用CAE软件进行模拟,获取训练样本,用Matlab语言编制应用程序,运用参数优化系统对浇口的位置进行优化计算,结合UG/KF,实现知识驱动.

The Optimization Design of Gate Location Based on Hybrid Neural Network and UG/KF

On the basis of analysis on the characteristic of genetic algorithm, an improved real-coded hybrid genetic algorithm was put forward on the disadvantages of slow convergence speed and premature convergence of the algorithm in the searching in the available space of parameters, and The geometric model was mainly introduced and achieved by knowledge fusion, an optimization design of gate location was established based on a hybrid neural network and UG/KF approach. The training-data was acquired through the CAE simulation, and the application program was written in Matlab engineering computing language. With the optimization system, the parameter for optimization of gate location was made. Combined with UG/KF, so that truly realized knowledge driven.