Research And Application Of Recursive Terminal Sliding Mode Approximate Optimal Control Strategy Based On Nonlinear System

In the optimal control problem based on nonlinear systems,the traditional classical control theory has obvious limitations.Adaptive Dynamic Programming(ADP)integrates the ideas of dynamic programming,reinforcement learning and artificial neural network,successfully solves the problems of traditional dynamic programming,avoids the dimension disaster,and shows strong advantages in solving nonlinear optimal control.The effective integration of adaptive dynamic programming control strategy and other control methods can better reflect its advantages and universality.In this paper,a recursive terminal sliding surface(RTSM)consisting of a fast nonsingular terminal sliding surface and a recursive integration terminal sliding surface is constructed.An adaptive dynamic programming(RTSM-ADP)control strategy based on a recursive terminal sliding surface is designed.The system has actuator saturation,resource constraints Model uncertainty,state is not completely known and so on.The main work and contributions of this paper include:(1)For nonlinear systems with input constraints and external disturbances,an adaptive dynamic programming control strategy based on recursive terminal sliding mode is proposed in this paper.The optimal consistency control problem of nonlinear systems is studied.Aiming at the problem of input constraints in engineering practice,a non quadratic function is introduced in the controller design to overcome the problem of saturated nonlinearity.(2)Based on the proposed control strategy,a new event triggering mechanism is designed.The proposed event triggering mechanism determines the update time of the neural network weight,and proves the overall stability of the system under the proposed triggering mechanism with Lyapunov method.The simulation results show that the triggering mechanism saves network communication resources.(3)For the n-order nonlinear system with unknown internal state,the modeling uncertainty,actuator fault and external interference are regarded as lumped uncertainty,and the lumped uncertainty is enhanced as a new state.The design problem of generalized fault observer is transformed into the design problem of state observer,and only the input and output are used to reconstruct the internal state of the system,and the extended state observer is established.A decentralized high-order terminal sliding mode controller based on the extended state observer is designed.The total uncertainty of the system can be directly compensated in real time during the control action.The extended state observer is eliminated in the controller to resist the lumped uncertainty,which effectively improves the response speed and control accuracy.The Lyapunov method is used to prove the convergence of the extended state observer and the overall stability of the system.(4)Based on the reconstruction of the internal state variables of the n-order system,the design of the ADP controller overcomes the fact that the design of the ADP controller depends on all the states of the system.In this paper,an approximate optimal compensation control scheme based on adaptive dynamic programming is designed using only the input and output of the system.The overall stability of the system is proved through the Lyapunov stability theory,and the experimental verification shows that the proposed algorithm has faster response speed and higher control accuracy.The basic structure of the algorithm proposed in this paper is a single evaluation neural network structure,which can not only maintain the convergence of neural network weights,but also directly obtain the control law by evaluating the weights of the neural network.In this way,the complexity of adaptive neural network design can be avoided,and the proof of the overall stability of the system is greatly simplified.For each problem and algorithm,this paper gives the stability analysis of the system.On this basis,the robot arm is taken as the research object,and the controller is designed according to the proposed method.Through simulation experiments and comparison,on the one hand,the effectiveness of the proposed control algorithm is proved;on the other hand,further analysis is made from the perspective of energy conservation and engineering realizability.Finally,based on the QNETVTOL2.0 vertical elevator system,the Lab VIEW control program and the overall control flow are designed for the control algorithm proposed in Chapter 3.The effectiveness of the proposed method in actual control is proved through physical experiments.