基于指令域大数据的零件尺寸预测方法
|
摘要
针对机械加工零件的尺寸误差人工检测效率低和在机检测成本过高的问题,提出了一种尺寸在线预测方法。基于指令域分析方法,通过在加工过程中获得的数控系统内部指令域大数据,结合实际加工参数与加工后测量的尺寸误差建立起非线性映射模型,该模型基于LM-BP神经网络与RBF神经网络学习实现零件尺寸误差的预测。最后进行对比研究,两种建模方法均能达到很好的预测效果。该方法适应加工参数发生变化的生产环境,能够对被加工零件尺寸误差进行自适应预测。
For the low efficiency of manual measurement of dimension error of machined parts and the high cost of OMI( On Machine Inspection),this paper presented a method of on-line prediction of dimension.The method uses instruction domain analysis method in NC machining process to study the changes in the size of the processed parts. This method combines the big data of instruction domain obtained during processing and actual machining parameters with size error of measurement and then build nonlinear model with these data. The model based on LM-BP neural network and RBF neural network learning can finish the prediction of parts size error. The methods are suitable for size error prediction under the changing processing environment.
For the low efficiency of manual measurement of dimension error of machined parts and the high cost of OMI( On Machine Inspection),this paper presented a method of on-line prediction of dimension.The method uses instruction domain analysis method in NC machining process to study the changes in the size of the processed parts. This method combines the big data of instruction domain obtained during processing and actual machining parameters with size error of measurement and then build nonlinear model with these data. The model based on LM-BP neural network and RBF neural network learning can finish the prediction of parts size error. The methods are suitable for size error prediction under the changing processing environment.
引文
[1]宾鸿赞.加工过程数控[M]. 2版.武汉:华中科技大学出版社,2004.
[2]李鹏飞,张琳娜,郑鹏,等.机械加工尺寸预报建模研究[J].机床与液压,2016,44(11):101-103.
[3]杨柳.针对典型加工的尺寸预报建模技术研[D].兰州:兰州理工大学,2016.
[4]Risbood K A,Dixit U S,Sahasrabudhe A D. Predic-tion ofsurface roughness and dimensional deviation by measuringcutting forces and vibrations in turning process[J]. Journalof Materials Processing Tech,2003,132(1):203-214.
[5]Mocnik D,Paulic M,Klancnik S,et al. Prediction of di-mensional deviation of work-piece using regression,ANNand PSO models in turning operation[J]. Tehnicki Vjesnik,2014,21(1):55-62.
[6]王灏宇,王涛,刘晓胜,等.基于多传感器数据融合的RBF神经网络加工尺寸预测[C]//中国仪器仪表学会精密机械分会99'精密工程学术讨论会,1999.
[7] Lera G,Pinzolas M. Neighborhood based Levenberg-Mar-quardt algorithm for neural network training[J]. IEEETransactions on Neural Networks,2002,13(5):1200-1203.
[8]孙倩.基于LM-BP神经网络的推荐算法的研究与应用[D].北京:北京交通大学,2016.
[9]吴宝强,孙炜,曹成.柔性和摩擦力不确定条件下RBF神经网络自适应轨迹跟踪方法[J].机械工程学报,2012,48(19):23-28.
[10]孙素芬,罗长寿.基于RBF神经网络的蔬菜价格预报研究[J].中国农学通报,2011,27(28):269-273.
[11]Chen J,Yang J,Zhou H,et al. CPS Modeling of CNCMachine Tool Work Processes Using an Instruction-DomainBased Approach[J]. Engineering,2015,1(2):247-260.
[2]李鹏飞,张琳娜,郑鹏,等.机械加工尺寸预报建模研究[J].机床与液压,2016,44(11):101-103.
[3]杨柳.针对典型加工的尺寸预报建模技术研[D].兰州:兰州理工大学,2016.
[4]Risbood K A,Dixit U S,Sahasrabudhe A D. Predic-tion ofsurface roughness and dimensional deviation by measuringcutting forces and vibrations in turning process[J]. Journalof Materials Processing Tech,2003,132(1):203-214.
[5]Mocnik D,Paulic M,Klancnik S,et al. Prediction of di-mensional deviation of work-piece using regression,ANNand PSO models in turning operation[J]. Tehnicki Vjesnik,2014,21(1):55-62.
[6]王灏宇,王涛,刘晓胜,等.基于多传感器数据融合的RBF神经网络加工尺寸预测[C]//中国仪器仪表学会精密机械分会99'精密工程学术讨论会,1999.
[7] Lera G,Pinzolas M. Neighborhood based Levenberg-Mar-quardt algorithm for neural network training[J]. IEEETransactions on Neural Networks,2002,13(5):1200-1203.
[8]孙倩.基于LM-BP神经网络的推荐算法的研究与应用[D].北京:北京交通大学,2016.
[9]吴宝强,孙炜,曹成.柔性和摩擦力不确定条件下RBF神经网络自适应轨迹跟踪方法[J].机械工程学报,2012,48(19):23-28.
[10]孙素芬,罗长寿.基于RBF神经网络的蔬菜价格预报研究[J].中国农学通报,2011,27(28):269-273.
[11]Chen J,Yang J,Zhou H,et al. CPS Modeling of CNCMachine Tool Work Processes Using an Instruction-DomainBased Approach[J]. Engineering,2015,1(2):247-260.