| Grinding and polishing robots have characteristics of high machining precision,good workpiece consistency,and high operation efficiency.They have widely been used for the surface treatment of complex workpieces with high-end equipment such as aerospace,rail transit,marine engineering and automotive electronics.Robots can be used to keep workers away from harsh and dangerous environments,and are the most representative application scenarios of “Replacing Humans with Machines”.Grinding and polishing robots based on teaching face challenges of machining workpieces with complex structure and shape,of various specifications and sizes,with rich,high-speed,and precise process procedures.To improve the level of robot grinding and polishing operations,it is necessary to develop a robot control system with both precision and flexibility.The research and development faces two technical problems:(1)how to realize accurate execution considering the weak hypothesis dependence of parameter design and excellent anti-interference performance under the conditions of model mismatch and multi-mode interference;(2)under the condition of multi-contact model matching and multi parameter coupling,how to realize active compliance control considering control structure switching and excellent transient performance.Focusing on these two key problems,this paper studies the key technologies and theories of modeling grinding and polishing robot dynamics and high-precision trajectory tracking,actuator constraint dynamics identification and adaptive compensation control,contact impedance relationship modeling and compliance control between robot and environment,contact force modeling and robust hybrid force/position control.This study lays a theoretical foundation for developing a high-performance grinding and polishing robot control system with both high-precision trajectory tracking and active compliance,and applies it to the surface grinding and polishing of automobile engine cylinder blocks to achieve good economic and social benefits.The main research contents and innovations of this paper are as follows.The dynamics of polishing robots are difficult to accurately obtain,which will lead to a significant reduction in the control accuracy of the model-based control method when the parameters are mismatched.To solve this problem,a robust adaptive trajectory tracking control method based on machine learning is proposed in this paper.The radial basis function neural network was used to identify each part of the system dynamics,including friction.In order to obtain a dynamic model of the robot,the segment identification technique reduced the complexity of a single network,and the network parameters were adjusted online through an adaptive method to obtain the optimal combination of network weight parameters.For possible external disturbances,identification errors and system dynamics that could not be incorporated into the identification model,the excellent robustness of sliding mode control was used to suppress them in order to improve the control accuracy.At the same time,in order to improve the dynamic performance of the system,a nonsingular terminal sliding mode surface was designed to ensure that the system can converge in finite time,improving the convergence speed of the system and avoiding the singularity of terminal sliding mode control.The unstructured working environment,complex workpiece structure and continuous contact with the grinding and polishing end causes inevitable actuator constraints,which will lead to the problem that the grinding and polishing robot cannot obtain the ideal control input.To solve this problem,an adaptive control method based on actuator constraint identification and compensation is proposed in this paper.In this method,two cascaded radial basis function neural networks are used to estimate and compensate the actuator dead time.The sliding mode control method is used to suppress the modeling uncertainty,external interference and joint friction,and the integral sli ding mode is used to weaken the chattering.Simulations and experiments showed that this method can correctly estimate the nonlinear characteristics of the dead zone,eliminate the influence of the actuator dead zone,and obtain satisfactory trajectory tracking performance.In addition,this method does not require the symmetry of a dead zone occurrence region and nonlinear function,and has good applicability.During continuous contact between the grinding and polishing end and the workpiece,small changes in displacement may lead to fluctuations of the contact force,which damage the actuator and workpiece.To solve this problem,an impedance control method based on the nonsingular terminal sliding mode was designed in the task space.Firstly,according to the impedance model parameters of the system,the given trajectory of the system was transformed into the executable impedance trajectory of the robot;then,an impedance control method based on nonsingular sliding mode was designed to track it.Control of the contact force is realized by adjusting the position or speed error.In addition,the exponential reaching law was adopted to reduce the chattering of sliding mode control and improve the quality of the dynamic response of the system.Aiming to control the instability caused by switching the traditional force position hybrid control structure,a force/position/direction hybrid control method of a non-rigid surface-polishing robot based on preset performance constraints and fuzzy neural networks is proposed in this paper.A unified error model of end position and contact force was established.Through the concept of designing an auxiliary velocity vector,the normal contact force was integrated with the z-direction displacement of the grinding and polishing end,so as to realize the simultaneous control of force and position and avoid the switching of the control structure.In addition,the rotation direction error is also considered in the error vector to ensure that the grinding and polishing device is always perpendicular to the workpiece surface.The specified performance function and the improved asymmetric obstacle Lyapunov function are used to transform the error and constrain it to the specific specified interval,so as to improve the dynamic response quality of the control system.For unmodeled dynamics and uncertainties of the system,a fuzzy neural network is used to identify and compensate them,so as to eliminate the adverse impacts on system performance.Combined with the project requirements of a large,domestic,automobile engine manufacturing enterprise for the automatic surface cleaning of engine block castings,as a team member,the relevant achievements and ideas of high-precision trajectory tracking and compliance control proposed in thi s paper can be integrated into the research and development of project technology and equipment.Designing and developing the control system and equipment of an automobile engine casting surface-cleaning robot consisted of:(1)Focusing on the problem of workpiece placement deviation in industrial field grinding and polishing operations,for which a workpiece position error measurement and correction system based on laser sensor was designed,adopting a onedimensional point laser sensor.After several measurements around the casting,the pose deviation relative to the standard model was calculated,and the feeding error was eliminated through a compensation algorithm;(2)Focusing on the problem that various potential errors on the operation site may lead t o waste products and could damage the robot,based on the idea of compliance control in Chapter 4 and Chapter 5,a rapid action protection system for the grinding and polishing robot was designed to realize the smooth,efficient and safe operation of the casting cleaning robot,ensuring the integrity and safety of the casting body structure,and that all indexes of the product meet the design requirements.The goal of “Replacing Humans with Machines” in harsh working environment has therefore been achieved,providing good economic and social benefits. |
PDF Full Text Request