深度强化学习的数控机床几何误差标定与补偿研究  

Evaluation Model of Man-machine Interface of CNC Machine Tools Based on Analytic Hierarchy Process and Neural Network

作　　者：李宁[1] LI Ning(Guangdong University of Petrochemicai Technology,Maoming 525000,China)

机构地区：[1]广东石油化工学院,广东茂名525000

出　　处：《机械制造与自动化》2021年第3期68-71,共4页Machine Building & Automation

摘　　要：为了提高数控机床的加工精度,需要进行数控机床的几何误差标定与补偿。采用自由来流与圆柱中心连线的准线性标定方法构建数控机床控制约束参数测量模型,进行数控机床的输出载荷计算和结构力学参数评估,通过特征值屈曲分析的方法进行数控机床的几何误差测量,采用深度化学习的方法进行数控机床几何误差测量和误差补偿控制。仿真结果表明,该方法提高了标定精度,误差补偿能力较强。To improve the machining accuracy of CNC machine tools,it is necessary to calibrate and compensate the geometric errors of CNC machine tools.A quasi linear calibration method of free flow and cylinder center line was used to construct the control constraint parameter measurement model of CNC machine tools.The output load of CNC machine tools and the evaluation of structural mechanical parameters were carried out.The geometric error of CNC machine tools was measured by eigenvalue buckling analysis,and the geometric error measurement and error compensation of CNC machine tools were Implemented by deep reinforcement learning method compensation.The simulation results show that the method has higher precision and better ability of error compensation.

关 键 词：深度强化学习 数控机床 几何误差 误差标定 参数测量 结构力学 

分 类 号：TH161[机械工程—机械制造及自动化]