复杂工程系统结构优化层次分解神经网络方法

A Hierarchic Decomposition Neural Networks Method for Complicated Engineering System Structure Optimization

复杂工程系统的设计通常涉及到大量各种类型的设计变量和约束。采用传统的整体优化方法求解这样一种大型设计问题,将是非常困难而耗时的。基于前馈神经网络的权重分析,提出一种基于神经网络的结构优化层次分解方法,较好地解决了这一问题。实例分析表明该方法是可行和有效的。

A complicated engineering system design usually comes down to a great lot kind of design variables aml constraints. It is verydifficult and time-consuming to solve this kind large design problem using the traditional whole optimization method. Based on weights analysis of feedforward neural networks, A hierarehic decomposition neural networks method for solving this problem is provided. Results of the case research show that the method is feasible and effective.