基于极限学习机的并联机器人无模型误差补偿算法研究
Study on Non-modeled Error Compensation Algorithm for Parallel Robots Based on Extreme Learning Machine
相铁武 1蒋欣怡 2高春晖 2左洪福 1乔贵方2
作者信息
- 1. 南京航空航天大学民航学院
- 2. 南京工程学院自动化学院
- 折叠
摘要
针对工业并联机器人精度性能限制其在精密装配等高端制造领域应用的问题,文中以六自由度Stewart并联机器人为研究对象,提出了一种基于极限学习机的无模型误差补偿算法,旨在有效提升六自由度Stewart并联机器人的位姿精度.首先,建立六自由度Stewart并联机器人坐标系框架.其次,提出基于极限学习机的无模型误差补偿算法,并通过实验分析确定了极限学习机的网络框架以及隐藏层神经元个数.实验结果表明,基于极限学习机的无模型误差补偿算法将六自由度Stewart并联机器人的最大位置和姿态误差分别有效降低了90.96%和 97.01%,精度提升效果优于传统的BP 神经网络和CSBP神经网络.
Abstract
Aiming at the problem that the precision performance of industrial parallel robot limits its application in high-end manufacturing fields such as precision assembly,taking a six-degree-of-freedom Stewart parallel robot as the research object,a model-free error compensation algorithm based on the extreme learning machine was proposed,which aimed to effectively improve the positional accuracy of the six-degree-of-freedom Stewart parallel robot.Firstly,the coordinate system framework of the six-de-gree-of-freedom Stewart parallel robot was established.Secondly,the model-free error compensation algorithm based on the ex-treme learning machine was proposed.The network framework of the extreme learning machine and the neuron number in the hidden layer were determined through experimental analysis.The experimental results show that the proposed model-free error compensation algorithm based on the extreme learning machine effectively reduces the maximal positional and pose errors of the six-degree-of-freedom Stewart parallel robot by 90.96%and 97.01%.The accuracy improvement is better than that of the tradi-tional BP neural system and CSBP neural network.
关键词
并联机器人/机器人标定/极限学习机/精度提升/误差补偿Key words
parallel robot/robot calibration/extreme learning machine/accuracy improvement/error compensation引用本文复制引用
出版年
2024
NETL
NSTL
万方数据