基于POE模型的工业机器人运动学参数二次辨识方法研究

针对工业机器人在高度制造领域精度不高的问题，本文提出了一种基于POE模型的工业机器人运动学参数二次辨识方法。阐述了基于指数积(Product of exponential，POE)模型的运动学误差模型构建方法，并建立基于POE误差模型的适应度函数；为实现高精度的参数辨识，提出了一种二次辨识方法，先利用改进灰狼优化算法(Improved grey wolf optimizer, IGWO)实现运动学参数误差的粗辨识，初步将Staubli TX60型机器人的平均位置误差和平均姿态误差分别从(0.648mm，0.212°)降低为(0.457mm，0.166°)；为进一步提高机器人的精度性能，再通过LM(Levenberg-Marquard)算法进行参数误差的精辨识，最终将Staubli TX60型机器人平均位置误差和平均姿态误差进一步降低为(0.237mm，0.063°)，机器人平均位置误差和平均姿态误差分别降低63.4%和70.2%。为了验证上述二次辨识方法的稳定性，随机选取5组辨识数据集和验证数据集进行POE误差模型的参数误差辨识，结果表明提出的二次辨识方法能够稳定、精确地辨识工业机器人运动学参数误差。

Quadratic Identification Method of Kinematic Parameters of Industrial Robots Based on POE Model

Aiming at the problem of insufficient precision performance of industrial robots in the high-end manufacturing field, a quadratic identification method of kinematic parameters of industrial robots based on POE model was proposed. Firstly, the construction method of the POE kinematic error model was presented. The fitness function based on the POE kinematic error model was established for kinematics identification. Secondly, a quadratic identification method was proposed to realize the parameter identification with high precision. At first, the improved grey wolf optimizer algorithm was applied to realize the primary identification of kinematic errors. The average comprehensive position error and average comprehensive attitude error of the Staubli TX60 robot were reduced from (0.648mm,0.212°) to (0.457mm,0.166°) respectively. In order to further improve the accuracy performance of the robot, the accurate identification of kinematic errors was carried out through the LM (Levenberg-Marquard) algorithm. The average comprehensive position error and average comprehensive attitude error of the Staubli TX60 robot were reduced to (0.237 mm, 0.063°). The average comprehensive position error and average comprehensive attitude error were reduced by 63.4% and 70.2%. Finally, in order to verify the stability of the above quadratic identification method, five different sets of identification datasets and validation datasets were randomly selected for the parameter error identification of the POE error model. The results showed that the proposed quadratic identification method was able to stably and accurately identify the kinematic parameter errors of industrial robots.