Research On Trajectory Tracking And Vibration Suppression For Rigid-Flexible Coupled Manipulator Systems Based On Robust Adaptive Learning Control

Rigid-flexible coupled manipulator systems(RFCMS)consist of a rigid link,a flexible link,and two corresponding joint drive motors.The RFCMS is a complex dynamic system with strong coupling,nonlinear and infinite dimensions,and has the characteristics of spatio-temporal variation.Compared to the traditional rigid manipulator,the RFCMS is because of lightweight,has strong adaptability,and has high human-computer interaction security,etc.The orbit in aviation maintenance,deep-sea exploration,and medical surgery and has been widely used in the exoskeleton robot,not only can it assist people to accomplish minimally invasive surgery and rehabilitation medical training,but Moreover,it can replace human beings with working in harmful environments such as radiation and high pressure,which broadens the way for scientific research and exploration in extreme environments such as deep-sea aviation,and becomes a research hotspot of new robots at present.However,when the flexible link of the RFCMS moves in a wide range,in addition to rigid deflection,it also produces multi-frequency elastic vibration and the two are coupled with each other,which is highly challenging to achieve accurate trajectory tracking control.In addition,the elastic vibration of flexible links not only reduces the control accuracy of equipment but also aggravates the wear of components,which becomes a big obstacle to practical engineering application,so the design of elastic vibration suppression law is particularly important.Therefore,this dissertation focuses on the dynamic modeling of the RFCM system and based on this,the study of robust adaptive learning control strategy with strong disturbance rejection,which can achieve accurate trajectory tracking and elastic vibration suppression of the RFCM system.Based on this,in line with the basic idea of from the RFCM system with small deformation to the RFCM system with large deformation,from full state feedback control to output feedback control,the main research work is as follows:(1)Two kinds of mathematical models of the RFCM system studied in this dissertation are determined.On the one hand,based on the elastic vibration description of the flexible link of the RFCM system and the classical system energy analysis mechanics,First,the equations of infinite-dimensional Partial differential equations(PDE)are derived by means of the Hamiltonian principle(also known as the virtual work principle)for RFCMS with small deformation moving in the vertical plane and complex terminal equations.Secondly,based on the assumed mode method and the corresponding boundary conditions,the infinite-dimensional partial differential model of the RFCM system with small deformation is reduced to the Ordinary differential equation(ODE)model of the RFCM system with small deformation in order to facilitate the design of the finitedimensional controller and the state observer.On the other hand,it is worth pointing out that,for the first time,this dissertation integrates the nonlinear flexible curvature and adopts the principle of polynomial expansion of the flexible deflection angle mode to analyze the energy of the RFCM system with large-deformation,and obtains the dynamic model of the RFCM system with large-deformation through the corresponding variation.Furthermore,in order to confirm the rationality of the modeling method adopted in this dissertation to study the large-deformation RFCM system,the dynamic model is transformed into a static model,and the model comparison is conducted with the existing statics model of large-deformation flexible link.(2)A robust adaptive learning state feedback control for infinite-dimensional RFCM system with small deformation and nonlinear constrained inputs is proposed.To perform at the end of the horizontal plane with the load of the small deformation RFCM system and the PDE model,respectively,considering the characteristics in the two kinds of mechanical input nonlinearity(input saturation and input backlash)and the end of the elastic vibration state under the limited,design a robust adaptive boundary based on disturbance observation iterative learning state feedback control strategy for trajectory tracking and vibration suppression,The composite energy function is constructed along the iteration axis to analyze the error convergence and vibration suppression of the closed-loop system.Then,for the infinite-dimensional PDE model of the RFCM system with small deformation moving in the vertical plane,the uncertainty of external disturbances is estimated by neural networks(NN)considering distributed disturbances,partial actuator failures,and unknown control directions.A robust adaptive fault-tolerant state feedback control law based on NN is designed,which not only ensures accurate trajectory tracking but also inhibits disturbance and elastic vibration.Finally,the effectiveness of the proposed robust adaptive learning state feedback control algorithm is verified by numerical simulations.(3)Robust adaptive learning output feedback control for finite-dimensional RFCMS with small deformation based on state observer is studied.For small deformation finite dimensional RFCM system with uncertain time delay and unknown system dynamics,trajectory tracking and vibration suppression under nonlinear backlash,additive and multiplicative actuator faults,and hysteresis input quantization are considered,respectively.In order to solve the problem that the position and velocity information of RFCMS is not easy to be accurately measured,a state observer based on Radial basis function Neural networks(RBFNN)was designed.At the same time,the filter saturation function with limited tracking error is introduced to restrict the range of output error.Then,based on the RBFNN state observer and filter function,a robust adaptive iterative learning output feedback controller is constructed to ensure the convergence of the tracking error and the learning of the first two elastic vibration modes of the finite-dimensional RFCM system with small deformation.(4)Finite-time robust adaptive learning state feedback control for large deformation RFCM system with uncertain dynamics is considered.On the one hand,for the unknown system dynamics of large deformation RFCM system,the minimum learning parameter(MLP)neural network is used to estimate and compensate,which reduces the computational complexity compared with RBF neural network.By equivalently transforming the controlled large deformation RFCM system into a tracking error dynamic system,a fixed time sliding mode surface is proposed,and a fixed time adaptive sliding mode trajectory tracking control based on MLP is designed.On the other hand,when the large-deformation RFCM system has unknown external disturbances,a nonlinear disturbance observer is designed,and a robust adaptive sliding mode trajectory tracking control law based on the disturbance observer is proposed based on the predetermined time stabilization function.In order to suppress the elastic vibration at the same time,a virtual controller and hybrid tracking trajectory are introduced,and a robust linear quadratic state feedback control law is designed.Under the joint action of the trajectory tracking control law tracking hybrid trajectory and virtual control,the trajectory tracking error and elastic vibration mode of the large-deformation RFCM system can be stabilized together.Finally,the effectiveness of the proposed robust adaptive learning state feedback control algorithm is verified by numerical simulation.