首页|基于PSO-LSSVR的机器人磨抛材料去除模型

基于PSO-LSSVR的机器人磨抛材料去除模型

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
  • 维普
为了建立磨抛工艺参数与材料去除深度的关系,建立一种基于最小二乘法支持向量回归机(LSSVR)的材料去除深度预测模型,并引入粒子群优化(PSO)算法来优化LSSVR的超参数,可提高LSSVR模型的预测准确性和全局优寻能力.搭建叶片机器人砂带磨抛实验平台,设计并进行多工艺参数实验,考虑工艺参数:砂带粒度、砂带转速、进给速度、接触力和叶片表面曲率半径,获得叶片表面的材料去除深度,最终利用实验数据建立了PSO-LSSVR叶片材料去除深度预测模型.结果表明,PSO-LSSVR模型的预测准确率为95.37%,平均预测误差为0.003 463,说明PSO-LSSVR模型具有较高的预测精度,并结合实际加工情况进行实验验证可行性,证明PSO-LSSVR模型可以有效合理地建立工艺参数与材料去除深度的关系.
Robot Polishing Material Removal Model Based on PSO-LSSVR
In order to establish the relationship between grinding process parameters and material removal depth,a material removal depth prediction model based on least square support vector regression(LSSVR)was first established,and particle swarm optimization(PSO)algorithm was introduced to optimize the LSSVR hyperparameters,which can improve the prediction accuracy and global optimization ability of the LSSVR model.The experimental platform of robot blade abrasive belt polishing was built,and multi-process parameter experiments were designed and carried out.The material removal depth of blade surface was obtained by considering process parameters such as sand belt particle size,sand belt speed,feed speed,contact force and blade surface curvature radius.Finally,the prediction model of PSO-LSSVR blade materi-al removal depth was established by using experimental data.The results show that:The prediction accuracy of the PSO-LSSVR model was 95.37%,and the average prediction error was 0.003 463,indicating that the PSO-LSSVR model had a high prediction accuracy.The feasibility was verified by experiments combined with the actual processing situation,which proved that the PSO-LSSVR model could effectively and reason-ably establish the relationship between process parameters and material removal depth.

robot abrasive belt grinding and polishingprediction modelprocess parametersleast squares support vector regressionparticle swarm optimization

蔡鸣、朱光、李论、赵吉宾、王奔

展开 >

沈阳航空航天大学机电工程学院,沈阳 110136

中国科学院沈阳自动化研究所,沈阳 110016

中国科学院机器人与智能制造创新研究院,沈阳 110169

机器人砂带磨抛 预测模型 工艺参数 最小二乘法支持向量回归机 粒子群算法

国家自然科学基金项目—辽宁省联合基金项目辽宁省自然科学基金计划面上项目

U19082302023-MS-034

2024

10.13462/j.cnki.mmtamt.2024.01.038

组合机床与自动化加工技术
大连组合机床研究所 中国机械工程学会生产工程分会

组合机床与自动化加工技术

CSTPCD北大核心
影响因子：0.671
ISSN：1001-2265
年,卷(期)：2024.(1)
  • 7