柔性关节双臂空间机器人的滑模神经网络控制

Sliding-mode Neural Network Control of Flexible-joint Dual-arm Space Robot

研究参数不确定、外部扰动影响下柔性关节双臂空间机器人的滑模神经网络控制问题。采用线性扭转弹簧等效描述关节弹性,根据系统线、角动量守恒原理及拉格朗日法,推导柔性关节双臂空间机器人的动力学模型。利用奇异摄动法对柔性补偿后的系统进行分解,得到两个可独立控制的子系统。为实现系统末端操作器在两种不确定性因素干扰下的轨迹精确追踪及关节振动抑制,提出一种刚性运动滑模神经网络控制、柔性运动反馈控制相结合的混合控制方案。仿真结果显示,所提方案具有较强的抗扰动及抗参数不确定能力,可使系统完成所期望的圆周运动并抑制关节的弹性振动。

The sliding-mode neural network control problem was discussed for a flexible-joint dual-arm space robot under the influences of parametric uncertainty and external disturbance. The elastic deformation of joint was described by the linear torsion spring, in terms of the principle of linear and angular momentum and the Lagrange method, the dynamic model of flexible-joint dual-arm space robot was derived. By using of the singular perturbation approach, two subsystems which can be controlled independently were separated from the system with the joint flexible compensation. To realize the accurately trajectory tracking of end-effectors on the effects of two uncertain factors and suppress the elastic vibration of joints, a hybrid control scheme composed by the sliding-mode neural network control of rigid motion and the feedback control of flexible motion was put forward. Simulation results show that the presented scheme has stronger ability to resist disturbance and parametric uncertainty, and can control the system to perform the desired circular motions and suppress the elastic vibration of joints.