Adaptive Control And Simulation Research For Robotic Systems With Unmodeled Dynamics

The manipulator system is a very complex multi input and multi output(MIMO)nonlinear system,and has many nonlinear dynamic characteristics,such as saturation,time delay and flexibility.In some occasions where the control accuracy of the manipulator is high,dynamic model of manipulator is needed.Due to the existence of external disturbances and measurement errors,the dynamic model is not very precise and complete.So when designing manipulator control algorithms,unmodeled dynamics must be considered.We can eliminate the effects of unmodeled dynamics by designing algorithms.On the other hand,in order to achieve effective control of the system,it is often necessary to constrain the input or output of the system.For the manipulator system,the input and output have upper and lower bounds.Once the system has crossed the boundary during operation,the stability of the system will be destroyed.Time-delay characteristics have also existed in the manipulator system.For systems with time-delay,the stability will be greatly affected.Therefore,for the manipulator system with unmodeled dynamics and output constraints,considering the time-varying delay and actuator dynamics,adaptive neural network control methods are designed by using the Barrier Lyapunov function.And in the Matlab environment,GUI simulation programs are written.The main contents of the thesis are as follows:Firstly,for a class of single-link flexible manipulator systems,considering the effects of unmodeled dynamics,adaptive control scheme is designed by using dynamic surface control methods;Radial basis function neural networks(RBF NNs)are used to approximate the unknown continuous nonlinear functions in the system;Introducing a dynamic signal to handle the uncertainties generated by unmodeled dynamics.Through theoretical analysis,it is proved that all signals in the closed-loop system are semi-globally uniformly ultimately bounded.Finally,the Matlab software is used to build the numerical simulation model of the manipulator,the simulation results illustrate the effectiveness of the proposed control scheme.Secondly,an adaptive neural network control scheme is proposed for a class of N-link flexible manipulator systems with unmodeled dynamics.Taking actuator dynamics into consideration,a DC motor is used to simulate the input of the manipulator,an auxiliary system is introduced to constrain the input voltage of the motor;Using the barrier Lyapunov function(BLF)to constrain the output of system;The block-structured RBF NNs are used to approximate the unknown nonlinear functions;Through theoretical analysis,it is proved that all signals in the closed-loop system are semi-globally uniformly ultimately bounded.,and both input and output are constrained within the specified range.Finally,a 2-DOF flexible manipulator is used to perform the numerical simulation and the results illustrate the effectiveness of the proposed control scheme.Thirdly,an adaptive control scheme was proposed for a class of N-link flexible manipulator systems with unmodeled dynamic and time-varying output constraints,considering the effects of time-delay,by choosing appropriate L-K functionals,unknown time-varying delay uncertainties were compensated;The adaptive controller designed by asymmetric barrier Lyapunov function(ABLF)carries out the time-varying constraint;Through theoretical analysis,it is proved that all signals in the closed-loop system are semi-globally uniformly ultimately bounded,and the output satisfies the time-varying output constraints.Finally,the 4-DOF flexible manipulator system was used for simulation and the results illustrate the effectiveness of the proposed control scheme.