Research On Finite/Fixed-time Control Of Robot System With Typical Nonlinearity

Robot systems are a complex class of highly non-linear,strongly coupled dynamical systems,and with the development of automation technology,their highprecision trajectory tracking control is one of the main focuses of robotics research.The frictional and viscous forces between the joints of the robot system and external disturbances make it difficult to obtain an accurate mathematical model.In addition,due to the limitations of the physical characteristics of the actuators,robotic systems often have non-linear characteristics at the input side,such as actuator failure,dead zones,saturation,quantization and other non-linearities.Furthermore,with increasing industrialisation and intelligence,new requirements are placed on the high quality control of robotic systems,such as fast convergence of errors.This paper explores finite/fixed time control schemes with the aim of compensating for the nonlinearity of the robotic system and enabling fast convergence of the system,with the following main work:1)For the trajectory tracking problem of uncertain robot systems containing input saturated nonlinearity,the hyperbolic tangent function is used to transform the asymmetric saturation into a symmetric saturation problem,and the joints of the robot are restricted to a specific range with the help of the barrier Lyapunov function;to improve the system response rate,a fractional power design finite time control method is introduced based on the traditional backstepping-based method and neural network control.2)For the trajectory tracking problem of uncertain robotic systems containing input dead zone nonlinearity,the dead zone nonlinearity is linearized using linearization techniques,followed by online approximation of the uncertain model of the robot and dead zone linearization error using adaptive neural networks,respectively;to further improve the convergence rate of the neural network and tracking error,the neural network weight update algorithm and robot control law are designed based on a fixed time mechanism in the framework of the inverse step method.3)For the uncertain robot system trajectory tracking problem containing input quantization nonlinearity,the linearization technique is used to linearize the quantization link.broad learning system is used to approximate the uncertain model of the robot system online,combined with the fixed time control method to design the neural network weight update law.A fixed time control scheme based on the broad learning system is proposed.4)For the trajectory tracking problem of uncertain robot system with actuator failure,an adaptive fixed-time fault-tolerant control strategy based on sliding mode control is designed.The neural network combined with the broad learning system is used to estimate the uncertainty of the system.The designed sliding mode function can effectively deal with actuator failure and avoid tremors.The advantage of this adaptive control method is to estimate the lower limit of the health status of the actuator when the health status of the actuator is unknown to avoid jumps in the Lyapunov function.