基于改进差分进化的3-RPS机器人逆运动学参数标定

Inverse Kinematic Parameters Calibration of 3-RPS Parallel Robot Based on Modified Differential Evolution

并联机器人结构较为复杂,由设计参数求解出的关节驱动不能使其达到理想位姿,然而并联机器人的位姿精度会直接影响到工作质量。为提高并联机器人逆运动学模型的精度,对所设计的3-RPS并联机器人建立逆运动学误差模型,并提出一种基于竞争的多变异策略差分进化算法对模型中的参数进行标定。在该算法中设计两个分别侧重局部开发和全局探索的种群,每个种群包含3种变异策略,并在每个种群中制定一种竞争制度用来挑选出标定过程中表现最好的策略,确保标定得到的参数最优。使用标定后的运动学参数对逆运动学模型进行修正,并通过Adams验证修正后模型的准确性。结果表明:所提算法收敛速度较差分进化算法提升了50%,且最终收敛值更小;与粒子群算法和差分进化算法相比,所提算法收敛结果不受进化过程中随机性带来的干扰;经该算法标定后,3-RPS并联机器人3个自由度精度较标定前分别提升了73.5%、88.7%、95.2%,精度得到显著提高。

Due to the complex structure of parallel robot,the joint torque solved by the design parameters cannot drive the parallel robot to achieve the ideal position and pose.However,the pose accuracy of parallel robot directly affects the work quality.To improve the model accuracy of the parallel robot,the kinematic error model was established for the designed 3-RPS parallel robot.In addition,on the basis of the traditional differential evolution(DE),a competitive multi-mutation differential evolution(CMDE)algorithm was proposed to calibrate the model parameters.In this algorithm,two populations were designed for the local exploitation and global exploration,where each population contained three mutation strategies.Moreover,a competitive system was developed in each population to select the better strategy in the calibration process,which could obtain the best optimal parameters.The kinematic parameters by calibration were used to modify the inverse kinematics model,and the accuracy of the modified model was verified by Adams software.The simulation results show that the proposed CMDE could achieve 50%faster convergence speed and smaller final convergence value in comparison to DE method.Also,compared with PSO and DE algorithms,the proposed CMDE had the strongest anti-interference ability in the evolution process.Moreover,compared with calibration before,the improvements of 3-RPS parallel robot with three degrees of freedom were 73.5%,88.7%and 95.2%,respectively.