三关节机器人的神经网络补偿自适应控制器设计

三关节机器人广泛用于工业生产、轮式或履带式排爆机器人,为了补偿由于机器人结构参数、作业环境干扰等不确定性因素造成的机器人动力学模型的不确定性,将机器人动力学模型分解为名义模型和误差模型两部分,其误差模型采用RBF神经网络进行补偿,得到其估计信息,神经网络的输出权值根据Lyapunov稳定性理论采用自适应算法进行调整。所设计的神经网络补偿自适应控制器解决了不确定性机器人动力学系统控制器设计的不确定性问题,同时,通过定义Lyapunov函数,证明了控制器能渐近、稳定地跟踪期望轨迹。机器人的3个关节在控制器的作用下,约在5 s时达到期望轨迹,神经网络约在5 s时逼近机器人动力学模型的误差模型,实验结果表明了机器人关节对期望轨迹具有良好的轨迹跟踪性能。

Design of Adaptive Controller Based on Neural Network Compensation for Robot with Three Joints

The robot with three joints are widely used to industrial manufacture and wheeled or crawler type explosive-handling robot. In order to compensate dynamic model’s modeling error caused by uncertain space manipulator’s structure parameter,interfere with the working environment and uncertain resonant mode for the space manipulator,the manipulator’s dynamic model is divided into nominal model and error model. The error model is compensated by RBF neural network. And the error model’s estimation information is obtained. The neural network’s output weights are adjusted by adaptive algorithm according to Lyapunov stability theory. Space manipulator’s neural network adaptive controller is solved the problem that controller’s design is uncertain for uncertain space manipulator’s dynamic system. Meanwhile,the controller can gradually and stably track desired trajectory though defining Lyapunov function. The controller is used to control joint’s torque for space manipulator with three joints. Three joints could trace the desired trajectory in 5 s. RBF neural network can approximate manipulator’s error model in 5 s. Simulation experiments test and verify manipulator’s joints have favorable desired trajectory tracking performance.