摘要
加工过程中为避免因机床异常振动造成的零件表面粗糙度突变,提出一种基于LSTM算法的表面粗糙度监控模型,通过对主轴与台面安装传感器,实现机床振动量的实时采集并作为时序变量输入模型。在M-V5CN组合机床铣削U71Mn高锰钢样本集上证明该模型可有效训练,且RMSprop优化器相对于Adam与SGD算法优化器可更有效降低模型泛化性误差。最终在嵌入式开发板中预测最大绝对误差低至0.01μm,平均误差为0.005μm可在加工中对表面粗糙度进行有效监控。
In order to avoid the sudden change of surface roughness caused by abnormal vibration of machine tool,a surface roughness monitoring model based on LSTM algorithm is proposed.By installing sensors on the spindle and table,the vibration of the machine tool can be collected in real time and input as time series variable.It is proved that the model can be effectively trained on the M-V5CN modular machine tool milling U71Mn high manganese steel sample set,and the rmsprop optimizer can reduce the model generalization error more effectively than Adam and SGD algorithm optimizers.Finally,the maximum absolute error is as low as O.O1μm and the average error is 0.005μm in the embedded development board,which can effectively monitor the surface roughness during machining.
作者
庄曙东
史柏迪
陈威
陈天翔
ZHUANG Shu-dong;SHI Bai-di;CHEN Wei;CHEN Tian-xiang(School of Mechanical Engineering,HoHai Univ.,Jiangsu Changzhou 213022,China;Jiangsu Key Laboratory of Precision Instruments,Nanjing University of Aeronautics and Astronautics,Jiangsu Nanjing 213009,China)
出处
《机械设计与制造》
北大核心
2023年第4期80-84,共5页
Machinery Design & Manufacture
基金
江苏省高校实验室研究会立项资助研究课题(GS2019YB18)
江苏省精密与微细制造技术重点实验室开放基金项目资助(JSKF201919)
中央高校基本科研业务费(2018B44614)。
