ART1神经网络在基于CBR概念设计中的应用  

Application of ART1 Neural Network in Conceptual Design Based on CBR

作　　者：薄瑞峰[1] 黄洪钟[1] 黄莉[2] 

机构地区：[1]大连理工大学精密与特种加工教育部重点实验室,辽宁大连116023 [2]中北大学机械工程系,太原030051

出　　处：《测试技术学报》2005年第4期447-452,共6页Journal of Test and Measurement Technology

基　　金：国家自然科学基金资助项目(59685003);高等学校全国优秀博士学位论文作者专项基金资助项目(200232)

摘　　要：提出一种机构编码方法,可以分别对机构的运动功能属性和功能质量属性进行二进制编码,在此基础上,深入探讨了ART 1神经网络在机构实例的分类和决策中的应用并对其算法进行了改进.与其它神经网络相比,ART 1网络的优点在于可以实现机构动态分类和决策,设计者可根据自己的意图,通过调节警戒参数得到满意或最优的设计方案.实例表明,该方法合理可行.A novel coding approach to mechanism is presented at first, by which the attributes of kinematic function and function quality for mechanism can be expressed respectively with a list of binary codes. Then based on an introduction to ART1 neural network, application of ART1 in classification and decision for mechanism cases is deeply investigated and some improvement for its learning algorithm is made. Compared with other neural network, the advantage of ART1 is that it can implement a dynamic classification and decision, viz. the designer can acquire an optimal design scheme or several satisfactory schemes in terms of his design intent by adjusting the value of vigilance parameter.

关 键 词：ART1网络 机构编码 动态分类决策 设计检索 CBR 机械运动方案设计 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]