基于RSM和GWO-BP混合代理模型的三维车削力传感器开孔位置多目标优化设计  

Multi-objective Optimization Design of Opening Position for 3D Turning Force Sensor Based on RSM and GWO-BP Hybrid Agent Model

作　　者：韩继科 王鹏 张昌明[1,2] 戴裕强 HAN Jike;WANG Peng;ZHANG Changming;DAI Yuqiang(School of Mechanical Engineering,Shaanxi University of Technology;Shaanxi Key Laboratory of Industrial Automation)

机构地区：[1]陕西理工大学机械工程学院 [2]陕西省工业自动化重点实验室

出　　处：《仪表技术与传感器》2024年第3期6-13,共8页Instrument Technique and Sensor

基　　金：陕西省秦创原科学家+工程师项目(2022KXJ-139);陕西省重点产业链项目(2023-ZDLGY-28);陕西省重点研发项目(2021GY-348)。

摘　　要：智能数控车床的研制需要配备智能化的切削力传感器,通过实时监控切削过程中切削力的变化,及时掌握工件和刀具的切削状态,文中针对十角环切削力传感器灵敏度不高(桥臂应力过小)的缺点,采用在环臂开通孔的方式提高局部应力进而得到高灵敏度的传感器。为得到最佳开孔位置,采用中心CL2偏差小(CL2=0.028)的最佳空间填充法对位置参数进行高维空间采样,对比2种模型各自的优势,以传感器的变形量、固有频率和路径应力为优化目标构造了RSM和GWO-BP的混合代理模型,对比不同算法的Pareto前沿、IGD和HV,确定选择SparseEA对混合代理模型进行多目标优化。优化后的传感器:变形量增加14.7%,等效应力增加155%,3个方向的灵敏度提升6倍左右。The development of an intelligent CNC lathe requires the integration of an intelligent cutting force sensor to effectively monitor the real-time changes in cutting force during the cutting process and promptly assess the cutting status of the workpiece and tool.In this paper,aiming at the shortcomings of the sensitivity of the ten-angle ring cutting force sensor(bridge arm stress is too small),an approach involving opening a hole in the ring arm was employed to enhance the local stress within the structure.To determine the optimal position for the opening,a high-dimensional parameter space was sampled using the optimal space filling method,with a small deviation of the center C_(L_(2)) value(C_(L_(2))=0.028).Compare the advantages of each of the two models,a hybrid agent model,combining RSM and GWO-BP,was constructed.The deformation,intrinsic frequency,and path stress of the sensor were considered as optimization objectives.To select the most suitable algorithm for multi-objective optimization of the hybrid agent model,the Pareto front,IGD,and HV of different algorithms were compared.SparseEA was chosen as the preferred algorithm for the multi-objective optimization.The optimized sensor exhibits a 14.7%increase in deformation and a significant 155% increase in local stress,about six times greater sensitivity in all three directions.

关 键 词：神经网络 PARETO前沿 车削力传感器 多目标优化 中心CL2偏差 

分 类 号：TH823[机械工程—仪器科学与技术]