移动机器人神经网络补偿计算力矩控制  

Computed Torque Control with Neural Network Compensation for Mobile Robot

作　　者：刘鑫 陈昌忠 罗淇 LIU Xin;CHEN Changzhong;LUO Qi(College of Automation&Information Engineering,Sichuan University of Science&Engineering,Zigong Sichuan 643000,China;Artificial Intelligence Key Laboratory of Sichuan Province,Zigong Sichuan 643000,China)

机构地区：[1]四川轻化工大学自动化与信息工程学院,四川自贡643000 [2]人工智能四川省重点实验室,四川自贡643000

出　　处：《机床与液压》2023年第11期52-58,共7页Machine Tool & Hydraulics

基　　金：四川省科技厅计划项目(20ZDYF0919);人工智能四川省重点实验室开放基金项目(2020RYJ05)。

摘　　要：针对存在动力学不确定建模项、建模误差及外界干扰的移动机器人,设计RBF神经网络补偿计算力矩控制算法。基于反步法设计运动学辅助速度控制率。根据动力学理想名义模型,基于计算力矩法设计一般的力矩控制器。在此基础上,建立具有不确定建模项、建模误差及外界干扰的移动机器人动力学模型,基于计算力矩法设计带有RBF神经网络补偿的力矩控制器,神经网络的权值由自适应律给出。最后,利用Lyapunov理论证明了系统的稳定性。仿真结果表明:神经网络对系统不确定项具有良好的逼近性能,相比于一般的计算力矩控制,所提出的神经网络补偿计算力矩控制算法具有更好的跟踪性能,控制系统具有更好的鲁棒性。For the mobile robot with dynamic uncertain modeling,modeling error and external disturbance,a computed torque control algorithm with RBF(radial basis function)neural network compensation was designed.The kinematic auxiliary speed control algorithm was designed based on the backstepping method.According to the dynamic ideal nominal model,a general torque controller was designed based on the computed torque method.On this basis,a dynamic model of the mobile robot with uncertain modeling,modeling error and external disturbance was established.The torque controller with RBF neural network compensation controller was designed based on the computed torque method.The weights of the neural network were given by the adaptive law.Finally,the stability of the system was proved by using Lyapunov theory.The simulation results show that neural network has good approximation performance for system uncertainty,compared with general computed torque control,the proposed computed torque control algorithm with RBF neural network compensation has better tracking performance,and the control system has better robustness.

关 键 词：移动机器人 RBF神经网络 计算力矩控制 LYAPUNOV理论 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]