Aiming at the common problems of unknown disturbance,nonlinear dynamics and parametric uncertainty in the electro-hydraulic servo system,a backstepping control(BC)method based on sliding mode observer and radial basis function(RBF)neural network was proposed.Firstly,a three-order strict feedback state space model of the valve-controlled cylinder system was constructed,and the sliding mode observer with finite time convergence characteristics was used to estimate the unknown speed state of the closed-loop system and the uncertain lumped disturbance of mismatched items.Then,the backstepping controller was designed based on the strict feedback model of the valve-controlled cylinder system and command filtered technology.The classic recursive backstepping methodology was utilized to design the nonlinear tracking controller according to the strictly feedback state-space mathematical model.The radial basis function neural network(RBFNN)was used to compensate the estimation error of the system speed and disturbance states,the filtering deviation and the matching term disturbance of the system,etc.According to Lyapunov stability theorem,the asymptotic bounded convergence of the tracking error for the closed-loop system was proved.Finally,valve-controlled cylinder electro-hydraulic control system test platform was established and the tracking performance of the control method on multi frequency reference signals was verified by comparative simulation and experiment.The research results demonstrate that the control method can further improve the trajectory tracking accuracy and robustness.Maximum experimental tracking error Me,average tracking error μ and mean square value of tracking error σ are respectively decreased by 66.7%,80%,and83.3%.Therefore,the proposed sliding mode disturbance observer and RBF compensator can effectively reduce the impact of various disturbances.
维普
万方数据