基于参数优化的机器人花键装配偏角感知识别方法  

Deflection angle perception and recognition method of robot spline assembly based on parameter optimization

作　　者：支乐威 陈教料[1] 王佳才 胥芳[1] 张立彬[1] ZHI Le-wei;CHEN Jiao-liao;WANG Jia-cai;XU Fang;ZHANG Li-bin(College of Mechanical Engineering,Zhejiang University of Technology,Hangzhou 310013,China)

机构地区：[1]浙江工业大学机械工程学院,浙江杭州310013

出　　处：《浙江大学学报（工学版）》2022年第3期452-461,共10页Journal of Zhejiang University：Engineering Science

基　　金：国家重点研发计划项目(2018YFC1309404);浙江省公益技术应用项目研究(LGG18E050023).

摘　　要：针对花键装配过程中存在轴孔偏角引起的卡阻导致装配成功率低的问题,提出基于参数优化的机器人花键装配偏角感知识别方法.根据花键装配的特点,利用力传感器采集花键装配过程中的力/力矩信号,基于混合鲸鱼优化算法(HWOA)的极限学习机(ELM)识别偏角的力信号并构建偏角经验库.结合支持向量数据描述(SVDD)算法,实现了未定义偏角的感知和偏角经验库的自我更新,以及用偏角的感知识别指导机器人完成花键装配任务.实验结果表明,所提方法对未定义偏角感知成功率和对已知偏角的识别精度分别为98.8%、98.12%,能有效指导机器人进行花键装配.A method for perception and recognizing the deflection angle of robot spline assembly based on parameter optimization was proposed,aiming at the problem of low assembly success rate due to the deflection angle of the shaft hole during the spline assembly process.According to the characteristics of spline assembly,the force sensor was used to collect the force/torque signal during the spline assembly process,an extreme learning machine(ELM)based on the hybrid whale optimization algorithm(HWOA)was used to identify the force signal of deflection angle and construct a deflection angle experience library.Combined with the support vector data description(SVDD)algorithm the perception of undefined deflection angle and self-updating of deflection angle experience library were realized,and the perception and recognition of deflection angle to guide the robot to complete the spline assembly task was achieved.The experimental results show that the proposed method has a success rate of 98.8%for sensing undefined deflection angles and 98.12%for recognizing known deflection angles,and can effectively guide the spline assembly.

关 键 词：机器人 花键装配 偏角感知识别 极限学习机 混合鲸鱼优化算法 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]