摘要
针对零件铣削加工变形难以在线检测的难点,提出了一种基于神经网络的加工变形在线预测方法,通过正交试验方法设计试验方案,进行了不同铣削参数条件下的铣削试验,以试验数据为训练样本建立了基于BP神经网络的铣削加工变形与铣削参数关系的预测模型。通过生产试验验证,此模型在样本参数覆盖范围内的模型精度可达99.56%、在覆盖范围外的模型精度高于95.47%,说明该神经模型能定量的反映出铣削参数与加工变形之间的关系。
In order to settle the troubles in on-line deformation measurement of workpiece,a deformation prediction method based on artificial neutral network was proposed.The cutting experiment scheme was designed by using orthogonal experiment technique.The experiments in different milling parameters were carried out,the workpieces' deformations were measured,and the deformation prediction model was created by using a Back-Propagation neural network and taking the experiment data as its training examples.According to the validation of the additional production experiments,the model's accuracy in the scope of examples is 99.56% and better than 95.47% outside the scope of examples.The results show that the deformation prediction model can accurately reflect the relationship between deformation and milling parameters.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2010年第8期1130-1133,共4页
Acta Armamentarii
基金
国防预先研究项目(513181604)
关键词
机械制造工艺与设备
变形
BP神经网络
铣削参数
预测模型
machinofacture technique and equipment
deformation
BP neural network
milling parameter
prediction model