基于FP算法的AVE局部模型划分  

Partition Method to Local Models of AVE Based on FP Algorithm

作　　者：赵立权[1] 方宏彬[1] 赵姝[1] 张铃[1] 

机构地区：[1]安徽大学智能计算与信号处理教育部重点实验室,安徽合肥230039

出　　处：《计算机技术与发展》2006年第6期168-170,共3页Computer Technology and Development

基　　金：国家自然科学基金资助项目(60475017);国家博士点基金资助项目(20040357002)

摘　　要：局部模型的划分是简化敏捷虚拟企业(Agile Virtual Enterprise,AVE)决策问题的重要方法,划分结果可能直接影响到敏捷虚拟企业的建立方式、AVE模型的优化以及合作伙伴的选择等。为了很好地完成AVE局部模型的划分问题,提出一种基于多层反馈神经网络的局部模型划分方法。该方法采用三层网络结构,各层完成指定的任务,第一层完成各输入向量的相似度是否大于给定值的判别;第二层则将相似度大于给定值的输入向量映射到同一类值,从而完成预分类;再利用反馈网络的反馈作用完成最后的分类。其结构清晰、分类灵活、学习的复杂性最低且网络的性能良好。Dividing the global model into local models is an important method of decomposing the decision problems of agile virtual enterprise （AVE）. The result of division may directly affect the way of setting up AVE, optimization of AVE model and member selection,etc. A method based on muhilayered neural networks with feedback connections was presented to solve this dividing procedure. The method adopts three- layered net structure, and each layer completes its appointed task： the first layer distinguishes whether similarity measure value of the input vector with others is larger than initial value or not; the second maps the two input vectors that their similarity value is larger than initial value to the same class, and makes classification in edvance;the feedback layer gets the last classification. The method has Clear net structure, flexible classification, the lowest complexity of learning and high performance of its net.

关 键 词：FP算法 敏捷虚拟企业 局部模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]