模糊神经网络在塑件注射成型多缺陷智能诊断中的应用研究  

Application of fuzzy neural network in intelligent multiple defects diagnosis for injection molding

作　　者：陈晨[1] 王鸿基[1] 徐迎强[1] 周结魁[1] 周毅 

机构地区：[1]合肥工业大学材料科学与工程学院,安徽合肥230009 [2]安徽毅昌科技有限公司,安徽合肥230601

出　　处：《模具工业》2012年第1期1-5,共5页Die & Mould Industry

基　　金：国家科技型中小企业技术创新基金(11C26213401942)

摘　　要：针对塑件注射成型多缺陷成因求解的模糊性与不确定性,考虑到神经网络在获取多维特征向量与对应输出向量之间非线性映射关系方面的优势,以及模糊技术在处理不精确信息方面的强大能力,提出将模糊理论和神经网络相结合,对塑件多缺陷成因进行判断、推理,并详细阐述了模糊神经网络用于塑件多缺陷诊断的整个过程。基于上述理论及Visual Prolog开发平台,开发了塑件注射成型多缺陷诊断智能系统,并进行了实例验证。结果表明,此系统具备较好的塑件多缺陷诊断能力以及一定的推广应用前景。In view of the ambiguity and uncertainty in solving the cause of injection molding,a method combining fuzzy theory and neural network for judging and determining the cause was proposed,which is because of the advantages of neural network in obtaining the non-linear mapping relationship between multidimensional eigenvector and output vector,and the strong ability of fuzzy technology in dealing with imprecise information.The whole process of the diagnosis was presented in detail.Based on the above theory and application platform of Visual Prolog,an intelligent system for the multiple defects diagnosis in injection molding was developed,and an application example was given.

关 键 词：注射成型 多缺陷 模糊神经网络 缺陷诊断 VISUAL PROLOG 智能系统 

分 类 号：TG76[金属学及工艺—刀具与模具] TQ320[化学工程—合成树脂塑料工业]