基于特征的飞机结构件数控加工工时预测模型  

Feature based man-hour forecasting model for aircraft structure parts NC machining

作　　者：刘长青[1] 李迎光[1] 王伟[1] 林勇[1,2] 

机构地区：[1]南京航空航天大学机电学院,江苏南京210016 [2]成都飞机工业(集团)有限责任公司,四川成都610092

出　　处：《计算机集成制造系统》2011年第10期2156-2162,共7页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(50905087);国家科技重大专项资助项目(2009ZX04014-051;2010ZX04015-011);航空科学基金资助项目(2010ZE52057)~~

摘　　要：针对目前数控加工工时预测方法不能兼顾精度和效率的问题,通过分析飞机结构件的结构特点和工艺特点,基于加工特征属性提炼工时的影响因素,提出了一种基于特征的两级结构工时预测模型。首先依据加工特征属性的数据类型把加工特征分为枚举型和数值型,然后以枚举型特征属性作为分类器的输入构建模型第一级结构,数值型特征属性作为反向传播神经网络的输入构建模型第二级结构。基于该模型开发的系统已经在某大型数控企业得到了良好的应用,效率高且误差在10%以内。Existing prediction methods for Numerical Control（NC） machining man-hours couldn＇t taken precision and efficiency into consideration simultaneously,by analzing the structure and process characteristics of aircraft structural parts,a feature-based two-level structure man-hour foresting model was proposed in view of man-hour influencing factors exteacted by machining feature attributes.Firstly,the machining features were classified into enumerative type and numerical value type according to machining feature attributes.Then,enumerative value type attribute was used as input of the classifier to establish first level of the model and numerical value type attribute was used as the input of back propagation neural network to set up second level of the model.The system developed based on this model was applied in a large NC enterprise with high efficiency and the error was within 10%.

关 键 词：飞机结构件 数控加工 工时预测 加工特征 反向传播神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]