Method of Industrial Robot Accuracy Compensation Based on Particle Swarm Optimization Neural Network

Abstract

Abstract:

For the absolute positioning accuracy of industrial robots used in aircraft flexible automated assembly cannot meet assembly precision,a neural network-based integrated accuracy compensation approach taking into account ambient temperature change factor was proposed based on robot spatial grid accuracy compensation method.Neural network's initial weights and thresholds were optimized by using particle swarm optimization method in order to prevent from falling into local minima in training.The experimental results show that the maximum value of the robot positioning error is as 0.32mm, and the mean value is as0.19mm with the temperature in the range of 20℃ to 30℃,which are much more better than the previous values 1~3mm,the absolute positioning accuracy can satisfy the requirements of aircraft automatic assembly.

Key words: industrial robot, accuracy compensation, neural network, calibration, absolute positioning accuracy

摘要：


针对工业机器人应用于飞机柔性化自动装配时绝对定位精度不能满足装配精度的问题,在机器人空间网格精度补偿方法的基础上,综合考虑环境温度的变化对机器人的绝对定位精度的影响,提出了基于神经网络的机器人综合精度补偿方法。为了防止神经网络在训练中陷入局部极值,采用粒子群优化方法对它的初始权值和阈值进行了优化。实验结果表明，当温度在20~30℃范围内变化时,机器人的绝对定位误差由补偿前的1~3mm,提高到补偿后的绝对定位误差最大值为0.32mm,平均值为0.194mm,精度较未补偿前有了大幅提高,可以满足飞机自动化装配的高精度的要求。



关键词: 工业机器人, 精度补偿, 神经网络, 标定, 绝对定位精度 

CLC Number:

TP242.2

ZHOU Wei-1, LIAO Wen-He-1, TIAN Wei-1, MO Shi-Meng-2, LIU Yong-2. Method of Industrial Robot Accuracy Compensation Based on Particle Swarm Optimization Neural Network[J]. China Mechanical Engineering, 2013, 24(02): 174-179.

周炜1, 廖文和1, 田威1, 万世明2, 刘勇2. 基于粒子群优化神经网络的机器人精度补偿方法研究[J]. 中国机械工程, 2013, 24(02): 174-179.

References

[1]许国康[1].大型飞机自动化装配技术[J].航空学报,2008,29(3):734-740. 6
[2]龚振邦等.机器人机械设计[M].北京：电子工业出版社,1995.. 5
[3]任永杰[1],邾继贵[1],杨学友[1],叶声华[1].利用激光跟踪仪对机器人进行标定的方法[J].机械工程学报,2007,43(9):195-200. 0
[4]焦国太[1],阿德·依科拉夫[2],余跃庆[1].工业机器人位姿误差的计算[J].机械科学与技术,2002,21(1):35-36. 1
[5]解则晓[1],辛少辉[1],李绪勇[1],徐尚[1].基于单目视觉的机器人标定方法[J].机械工程学报,2011,47(5):35-39. 6
[6]Judd R P,Knasindki A B. A Technique to CalibrateIndustrial Robots with Experimental Verification[J]. IEEE Transaction on Robotics and Automa-tion, 1990,6(1):20-30.
[7]蔡鹤皋,张超群.机器人实际几何参数识别与仿真[J].中国机械工程,1998,9(10):11-14. 9
[8]Xu W L,Wurst K H,Watanabe T,et al. Calibrationa Modular Robotic Joint Using Neural Network Ap-proach[C]//Proceedings of IEEE International Con-ference on Neural Networks. NewYork: IEEE,1994:2720-2725.
[9]Zhong X L, Lewis J M, Rea H. Neuro— AccuracyCompensator for Industrial Robots [C]//Proceed-ings of IEEE International Conference on NeuralNetworks. NewYork:IEEE,1994 :2797-2802.
[10]夏凯,陈崇端.补偿机器人定位误差的神经网络[J].机器人,1995,17(3):171-176. 0
[11]周炜[1],廖文和[1],田威[1],李东明[2],刘勇[2].面向飞机自动化装配的机器人空间网格精度补偿方法研究[J].中国机械工程,2012,23(19):2306-2311.
[12]朱求安[1],张万昌[1],余钧辉[1].基于GIS的空间插值方法研究[J].江西师范大学学报：自然科学版,2004,28(2):183-188. 08
[13]林忠辉,莫兴国,李宏轩,李海滨.中国陆地区域气象要素的空间插值[J].地理学报,2002,57(1):47-56. 82
[14]曲巍崴[1],董辉跃[1],柯映林[1].机器人辅助飞机装配制孔中位姿精度补偿技术[J].航空学报,2011,32(10):1951-1960. 2
[15]张文增[1],陈强[1],孙振国[1],汤晓华[1].弧焊机器人工件坐标系快速标定方法[J].焊接学报,2005,26(7):1-4.