基于RBF神经网络自适应控制的下肢外骨骼步态跟踪

The gait tracking of lower limb exoskeleton based on RBF network adaptive control

针对开发研制的下肢外骨骼机器人控制策略的需要,提出一种基于RBF自适应控制的外骨骼控制方法.建立了关于外骨骼的动力学模型,采用RBF网络分别实现对下肢外骨骼模型动力学方程中的重力项、哥氏力及离心项、正定惯性矩阵的逼近建模;通过实验获取髋关节与膝关节于步行过程中的数据,实现了曲线的拟合并将其作为理想输入,通过对比PID、RBF控制方法去控制外骨骼逼近步态曲线.由扰动前后的效果对比可知,基于RBF神经网络自适应控制算法的外骨骼平台可以跟踪步态轨迹,有利于提高系统对位置和速度的跟踪能力以及系统的稳定性.

Aiming at the needs of the developed lower limb exoskeleton robot control strategy,this paper proposes an exoskeleton control method based on RBF adaptive control.The dynamic model of the exoskeleton was established,and the RBF network was used to realize the approximation modeling of the positive definite inertia matrix,the centrifugal force and the Coriolis force term,and the gravity term in the dynamic equation of the lower limb exoskeleton model.The hips during walking were obtained through experiments.The data of joints and knee joints are fitted with curves and used as ideal input.By comparing PID and RBF control methods to control the exoskeleton to approach the gait curve,the comparison before and after adding disturbances shows that it is adaptive based on RBF neural network.The exoskeleton platform of the control algorithm can realize the tracking of the gait trajectory,which is beneficial to improve the system's ability to track position and speed and the robustness of the system.