液压伺服位置系统的神经网络backstepping控制  

Neural network backstepping control of hydraulic servo position system

作　　者：方一鸣[1,2] 李叶红[1] 石胜利[1] 李建雄[1] 

机构地区：[1]燕山大学工业计算机控制工程河北省重点实验室,河北秦皇岛066004 [2]国家冷轧板带装备及工艺工程技术研究中心,河北秦皇岛066004

出　　处：《电机与控制学报》2014年第6期108-115,共8页Electric Machines and Control

基　　金：国家自然科学基金(61074099);河北省高等学校创新团队领军人才培育计划(LJRC013);秦皇岛市科学技术研究与发展计划(2012021A006)

摘　　要：针对液压伺服位置系统存在的参数不确定性、外部干扰和输入饱和的问题,提出了一种神经网络backstepping控制算法。设计了神经网络辅助状态观测系统,并根据辅助状态观测误差来调节神经网络的权值,进而实现对系统复合干扰的在线观测。把该复合干扰的观测值引入到backstepping控制设计中,使得控制器能够对系统的复合干扰进行有效补偿;在backstepping设计过程中采用二阶滑模滤波器以避免微分项爆炸问题,简化了控制器的设计。通过Lyapunov稳定性理论证明了闭环系统所有信号一致最终有界。仿真结果表明,所设计的控制器能够有效地削弱参数不确定性、外部干扰和输入饱和对系统的影响,增强了系统的鲁棒性,实现了系统输出对期望位置的准确跟踪。A neural networks backstepping control scheme was proposed for the hydraulic servo position system with parameter uncertainties,external disturbance and input saturation problem.Firstly,the auxiliary state observer system was designed by the neural network,and the weights of neural network were adjusted according to the auxiliary state observation error to realize the online observation of compound disturbance of the system.Secondly,the observyation of compound disturbance was introduced into the backstepping controller design,which enables the controller to compensate effectively the compound disturbance of the system.The design of backstepping controller was simplified because a second-order sliding mode filter was utilized to avoid the problem of the explosion of terms.The Lyapunov theory was used to guarantee that all the signals in the system are uniformly ultimately bounded.Simulation results show the designed controller effectively weakens the influence of parameter uncertainties,external disturbance and input saturations,enhances the robustness of the system and makes the system output track accurately the desired position.

关 键 词：液压伺服系统 神经网络 干扰观测器 BACKSTEPPING控制 输入饱和 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]