基于可变异混合PSO的上肢康复机器人轨迹优化

Upper limb rehabilitation robot trajectory optimization based on variable mutation hybrid particle swarm optimizationrch

针对上肢康复机器人的训练轨迹运动生涩、不贴合人手臂的日常习惯等问题,提出使用改进粒子群算法对上肢康复机器人进行时间最优轨迹优化.由于粒子群算法极易早熟,存在寻优稳定性较差的问题,在算法中引入鸟类混群协同觅食的概念,设计了一种新的速度更新公式,考虑个体极值、群体极值和社会极值对粒子更新的影响,提高了算法在高维搜索空间中和高维约束条件下的全局寻优能力与寻优鲁棒性,并在粒子迭代过程中引入停滞变异补偿机制增加粒子多样性,进一步提升算法的寻优效率.通过MATLAB仿真实验,改进后的粒子群算法将平均最优时间缩短44.29%,平均收敛代数减小了16.75%,使得上肢康复机器人各关节运动符合康复训练要求,能够提高康复训练质量.

Regarding the cumbersome motion trajectories of upper limb rehabilitation robots,which do not align with the daily habits of human arms,we propose the use of an improved particle swarm algorithm for time-optimal trajectory optimization.Addressing the issue of premature convergence and poor optimization stability inherent in the particle swarm algorithm,we incorporate the concept of flock foraging coordination in birds into the algorithm.A novel velocity updating formula is designed,taking into account the influences of individual extremum,group extremum,and social extremum on particle updates.This design enhances the algorithm’s global optimization capability and optimization robustness in high-dimensional search spaces and under high-dimensional constraint conditions.Additionally,a stagnation mutation compensation mechanism is introduced during particle iteration to increase particle diversity,further improving the algorithm’s optimization efficiency.Through MATLAB simulation experiments,the improved particle swarm algorithm reduces the average optimal time by 44.29%and decreases the average convergence iterations by 16.75%.Furthermore,the joint movements of the upper limb rehabilitation robot conform to the requirements of rehabilitation training,thereby enhancing the quality of rehabilitation training.