摩托车智能设计实例推理系统的索引模型

为提高实例推理系统中实例检索的效率与质量，提出了基于人工神经网络的实例检索模型。该模型在实例层次组织的基础上，利用自适应共振网络实现对实例的动态分类，以缩小实例搜索的范围。采用前馈型神经网络记忆各实例的索引，以便在缩小的实例范围内快速地提取相似实例，提高检索效率与质量。最后给出了摩托车总体设计过程方案选择的算例，并将检索的结果与基于最近邻法的实例检索结果进行比较分析，结果证明了该模型的有效性。

Indexing model of case-based reasoning system for motorcycle intelligent design

To improve efficiency and quality of case retrieval in cased-based reasoning system,a case retrieval model based on the artificial neural network was presented.Firstly,based on cases hierarchy,dynamic classification of cases was conducted through self-adaptive resonance theory neural network so as to shrink the case-searching range.Secondly,the back propagation neural network was applied to memorize the indexes of the clustered cases. Thus the similar cases could be retrieved quickly in the shrinked case base and the retrieval efficiency and quality were improved.Finally,some examples of plan selection of motorcycle overall design were given to illustrate the working process of the system.Compared the retrieval results based on neural network with those based on nearest neighbor method,the effectiveness of case retrieval model was proved.