摘要
针对传统摩托车设计过程中,设计人员为了找出合乎要求的相似车型,过多依赖经验,主观确定车型属性参数权重的问题,将人工神经网络理论和算法应用于摩托车实例库检索系统的研究中。通过ART1神经网络实现摩托车实例库的动态聚类,形成聚类模板,再利用前馈(BP)神经网络对模板中的每一实例进行检索。这种方法减小了车型属性权重中主观因素的影响,提高了系统的检索质量与效果,在实际应用中有良好的前景。
To solve the problem that designers rely more on experience and decide the parameter weight of motorcycle subjectively for finding the similar motorcycle case in the traditional motor design process, is written. The theory and the algorithm of artificial neural network are applied in motorcycle case base retrieval system. The ART1 neural network is used to cluster the motorcycle case base and form the clustered template. The BP neural network is used to retrieve cases in the clustered template. According to this, the effect of subjective factor in the process of deciding property weight is lessened, the retrieval efficiency and quality is enhanced. It has a bright application future.
出处
《现代制造工程》
CSCD
2007年第10期8-11,共4页
Modern Manufacturing Engineering
基金
国家自然科学基金项目(50375156)
重庆科技攻关资助项目(7823-10)
