基于BP神经网络补偿的下肢外骨骼位置控制

Position Control Based on BP Neural Network Compensation for Lower Limb Exoskeleton Rehabilitation

坐卧式康复外骨骼机器人可以帮助下肢功能障碍患者进行康复训练,为实现康复训练的位置控制目标,通过深入分析人机耦合关系,建立包括穿戴者、外骨骼在内的耦合运动学/动力学模型,提出了基于人机混合系统动力学模型的带前馈补偿项的PD控制算法,然而为了实现良好的运动轨迹跟踪性能,必须在位置控制算法中加入对不确定项的补偿,因此提出一种基于BP神经网络补偿的位置控制算法并通过仿真验证算法的高效性。

Sitting/lying rehabilitation exoskeleton robot enable patients with lower limb dysfunction to take rehabilitation training,in order to achieve the position control goal of rehabilitation training,a coupled kinematics/kinetics model including wearer and exoskeleton was established by deeply analyzing the human-machine coupling relationship in the paper.A PD control algorithm with feedforward compensation term was proposed based on the dynamic model of human-machine hybrid system.However,in order to achieve good trajectory tracking performance,it is necessary to add uncertainty to the position control algorithm.The paper proposeD a position control algorithm based on BP neural network compensation whose validity can be verified by simulation.