【摘要】 针对摩托车总体设计过程依据相似性进行局部创新设计的特点,提出了基于CBR与神经网络的摩托车智能设计系统结构。采用面向对象方法对摩托车实例知识进行表示与组织,并将神经网络引入到实例推理中,利用ART1网络实现基于特征的动态层次聚类,通过BP网络实现从聚类模板到每一实例的索引,从而提高系统的推理效率与质量,形成有效的推理系统。更多还原

【Abstract】 In general,the overall design of the motorcycle is an innovative design in local based on the similarity of the mature product.According to this,the intelligent design system based on CBR and neural network is presented.The cases of the motorcycle are expressed and organized with the object-oriented method.Meanwhile,the neural network is introduced into the CBR system.ART1 neural network is used to cluster dynamically the cases based on the features and BP neural network is applied to index each case based on the clustered template.Therefore,the reasoning efficiency of the CBR system is enhanced and an effective reasoning system is achieved.更多还原