Research On Robot Robust Control Method Optimized By Convolutional Neural Network

Robots have a wide range of applications in the manufacturing industry,especially industrial robots,which are widely used in automobile assembly,production packaging,product testing,quality inspection,spray painting and welding.Among the many research directions of industrial robots,trajectory tracking control is an important direction.With the continuous development of technology,modern industrial robot control requires that while accelerating the response speed,improving the accuracy of trajectory tracking and improving product quality.Aiming at the problem of low accuracy and slow convergence in the trajectory of the control reference object caused by dynamic modeling errors and external uncertain interference in robot control,this paper builds on the basis of three typical robust control methods.The convolutional neural network based on deep learning optimizes the robust control algorithm,thereby improving the control accuracy and speeding up the trajectory convergence speed.The research content mainly includes:Aiming at the traditional robust variable structure control method,a convolutional neural network is introduced for optimization,and a convolutional neural network sliding mode variable structure control method is proposed.A convolutional neural network based on deep learning is constructed to compensate the robot's modeling errors and external uncertain interference parts,and proves the stability of the control system.The simulation results show that the optimized convolutional neural network sliding mode variable structure control method can solve the"chattering"problem,improve the tracking accuracy of the trajectory,and speed up the response speed.Aiming at the problem of the uncertain upper bound part in the traditional robust control method,the convolutional neural network model is introduced to compensate the uncertain upper bound part,which solves the problem of generally low steady-state accuracy of the traditional robust control.The simulation results show that when the optimized control method is used to complete the robot trajectory tracking control task,the convergence of the trajectory tracking error is accelerated and the control accuracy is improved.Aiming at the traditional robust H~? control method,the evaluation signal of the robot control system is first designed,and then the convolutional neural network model is introduced to optimize the robust H~? control method.Finally,the stability of the control system is proved through the HJI inequality.Compared with the robust H~? control method based on RBF neural network,the experimental results show that the optimized control method reduces the robot trajectory tracking error and accelerates the convergence speed of trajectory tracking.In this paper,the MATLAB-simulink module is used as the simulation platform,and the two-joint robot is used as the simulation object.The trajectory tracking of three robust control methods optimized by convolutional neural network is simulated,and good control effects have been achieved.However,the method optimized in this paper is not limited to use in two-joint robots as an optimization algorithm,and can also be extended to Multi-joint robots.