Hopfield神经网络与专家系统相结合的换热网络设计方法有效性分析

Efficiency Analysis of HNN/ES Based System for Heat Exchanger Network Design

在利用Ｈｏｐｆｉｅｌｄ神经网络（ＨＮＮ）模型搜索换热网络中最佳物流匹配方式时，通过建立专家系统来描述匹配过程中必须考虑的定性及定量因素，协助调整ＨＮＮ模型运行过程中的寻优方向，以减小ＨＮＮ模型的搜索空间，保证问题的全局优解。文章通过实例说明了该混合方法的有效性。

A Combined approach is presented for heat exchanger network synthesis in which expert system(ES) is used to aid the Hopfield neural network(HNN) in optimum searching. The efficiency of the approach is proved by examples. It shows that with the aid of ES, the dependency of HNN model on parameters is reduced and the global optimum is guaranteed. Furthermore, the method is practical in solving complex problems within reasonable time.