机械臂动力学参数的多变异策略花授粉算法辨识

Manipulator Dynamics Parameters Identification Based on Multiple Mutation Strategies Flower Pollination Algorithm

为了准确辨识出机械臂动力学参数,提出了多变异策略花授粉算法的参数辨识方法。分析了PUMA560机械臂结构和工作原理,使用改进的牛顿-欧拉方程建立了驱动力矩与动力学参数的函数关系。以驱动力矩反算值与实际值的误差最小为目标,建立了动力学参数优化模型。从原理层面分析了花授粉算法在迭代后期进化能力弱、易陷入局部最优的原因,提出了带随机机制的定向变异策略和均匀变异策略,将两种变异策略融入到花授粉算法中,得到了多变异策略花授粉算法。使用多变异策略花授粉求解优化模型,经实验验证,从辨识精度、力矩反算值与实际值相关性等两个角度讲,多变异策略花授粉算法的辨识结果远优于花授粉算法,证明了多变异策略花授粉算法在参数辨识中的有效性。

To identify manipulator dynamics parameters accurately,the parameters identification method based on multiple mutation strategies flower pollination algorithm is proposed.Structure and working principle of PUMA560 is analyzed.The function relationship of driving moment and dynamics parameters is built by improved Newton-Euler equation.Aimed at minimizing error of driving moment calculated value and real value,optimization model of dynamics parameters is built.The reason of weak evolutionary ability and easy to fall into local optimization is analyzed in principle.Orienteering mutation strategy with random mechanism and even mutation strategy are put forwarded.Introducing the two mutation strategies into flower pollination algorithm,multiple mutation strategies flower pollination algorithm is raised.Using the algorithm to solve optimization model,it is clarified that identification result of multiple mutation strategies flower pollination algorithm is optimal to traditional algorithm from the two aspects of identification accuracy and relevance of torque moment calculated value and real value.The result above proves the validity of multiple mutation strategies flower pollination algorithm on parameters identification.