基于改进神经网络算法的螺杆砂带磨削表面粗糙度预测研究  

Surface Roughness Prediction of Screw Belt Grinding Based on Improved Neural Network Algorithm

作　　者：董浩生 杨赫然[1,2] 孙兴伟[1,2] 董祉序 刘寅 DONG Hao-sheng;YANG He-ran;SUN Xing-wei;DONG Zhi-xu;LIU Yin(College of Mechanical Engineering,Shenyang University of Technology,Shenyang 110870,China;Key Laboratory of Numerical Control Manufacturing Technology for Complex Surfaces of Liaoning Province,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]沈阳工业大学机械工程学院,沈阳110870 [2]沈阳工业大学辽宁省复杂曲面数控制造技术重点实验室,沈阳110870

出　　处：《表面技术》2022年第4期275-283,共9页Surface Technology

基　　金：辽宁省自然科学基金指导计划(2019-ZD-0206);辽宁省“兴辽英才计划”(XLYC1905003);中央引导地方科技发展专项资金(2020JH6/10500048);国家自然科学基金(52005347);辽宁省博士启动项目(2019BS181)。

摘　　要：目的探究工艺参数对螺杆转子砂带磨削表面质量的影响规律。方法采用工件轴向进给速度为100~300 mm/min、砂带线速度为4.4~13.1 m/s、砂带张紧压力为0.2~0.3 MPa、磨削压力为0.4~0.5 MPa、砂带粒度为120~800目的工艺参数进行螺杆转子砂带磨削正交实验,基于改进的神经网络算法,建立螺杆转子砂带磨削后的表面粗糙度值预测模型,对磨削后的工件表面质量进行预测及分析。在此基础上采用预测模型针对各工艺参数对磨削质量的影响规律进行预测分析。结果利用多因素磨削实验获得预测样本及对比样本,对比结果表明,预测模型的平均训练精度约为93.38%,预测精度为92.46%。螺杆转子砂带磨削表面粗糙度值的单因素预测结果表明,工件表面粗糙度值随着接触轮正压力及磨削装置轴向进给速度的增加而升高,随着砂带线速度及砂带粒度的增加而降低。结论提出的算法可为选择螺杆转子砂带磨削的工艺参数提供理论依据。适当提高砂带线速度及砂带粒度、降低接触轮气缸压力及磨削装置轴向进给速度可获得较高的表面质量。This paper aims to explore the influence of process parameters on the surface quality of screw rotor abrasive belt grinding.The orthogonal experiment of screw rotor abrasive belt grinding is carried out for the axial feed speed of workpiece is 100~300 mm/min,the linear speed of abrasive belt is 4.4~13.1 m/s,the tension pressure of abrasive belt is 0.2~0.3 MPa,the grinding pressure is 0.4~0.5 MPa and the mesh number of abrasive belt is 120~800.Based on the improved neural network algorithm,the prediction model of surface roughness value after screw rotor abrasive belt grinding is established to predict and analyze the surface quality of the workpiece after grinding.On this basis,the influence of process parameters on grinding quality is predicted and analyzed by using the prediction model.Using multi-factor grinding experiments to obtain prediction samples and comparison samples,the comparison results show that the average training accuracy of the prediction model is about 93.38%and the prediction accuracy is 92.46%.The single factor prediction results of screw rotor abrasive belt grinding surface roughness value show that the workpiece surface roughness value increases with the increase of contact wheel positive pressure and axial feed speed of the grinding device,and decreases with the increase of abrasive belt linear speed and abrasive belt mesh.It can be seen from the above research results,the proposed algorithm can provide a theoretical basis for the selection of process parameters of screw rotor abrasive belt grinding.Higher surface quality can be obtained by appropriately increasing the linear speed and mesh number of the abrasive belts,reducing the cylinder pressure of the contact wheel and the axial feed speed of the grinding device.

关 键 词：磨削 表面粗糙度 麻雀搜索算法 神经网络预测 

分 类 号：TH161[机械工程—机械制造及自动化]