Neural Network Based Active Disturbance Rejection Control Of Hysteresis Nonlinear Systems

In this paper,a series of adaptive control schemes based on neural network active disturbance rejection are proposed for a class of non-affine nonlinear systems with unknown hysteretic disturbances and uncertain nonlinearity caused by the deviation of control parameters from nominal values.The proposed control scheme is applied to the metal cutting machine system.The stability of the system is discussed and analyzed.The strict mathematical proof and simulation verification are given.The main innovations are summarized as follows:A single loop hysteresis loop is constructed by using sinusoidal signal and backlash hysteresis operator.RBFNN and ADRC are combined to derive the controller expression according to differential implicit.A new dual channel composite controller design scheme is proposed by using reverse step design technology.By using the approximation of neural network,the compound controller overcomes the disadvantage of low control accuracy of traditional ADRC,and inherits the advantages of estimating and compensating unknown disturbance of the system without detailed model.Theoretical analysis and simulation results verify the effectiveness of RBF-ADRC algorithm.A multi loop hysteresis loop is constructed by using the control signal and the backlash hysteresis operator.By using feedback linearization technology,BP neural network and linear ADRC are combined to solve the problems of difficult parameters tunning,and the shortcomings of the bandwidth limited.It is more conducive to the application of the composite control based on the ADRC framework in the actual project development.Theoretical analysis and simulation verify the effectiveness and applicability of bp-ladrc control algorithm.The adaptive law of neural network weight and the adaptive law of ADRC gain are designed,and the detailed stability analysis of the compound controller is carried out,which provides a new idea for the stability analysis of the combination of ADRC and other control methods.In order to verify the robustness of compound control and its applicability in practical application more effectively,random noise interferenceand sinusoidal interference are added to the controller while the controller is unchanged.The simulation results show that the robustness of the composite controller is satisfactory.Based on the characteristics of non-linear,time-delay and hysteresis of the metal cutting machine system caused by chattering,the neural network active disturbance rejection control is applied to the metal cutting machine system.The controller of the metal cutting machine system is designed by using the control scheme proposed in this paper.Through the comparison with the traditional control scheme and the simulation verification of robustness,it is shown that the proposed control algorithm has strong robustness and can effectively suppress chattering.