Research On Transmission Error Control Based On Data-driven Method And Networked Control

As a component that transmits motion and force,the mechanical transmission system is an important part of CNC machine tools.However,due to the influence of multiple internal and external factors,transmission error occurs in the process of mechanical transmission,which would lead to the failure of CNC machine tool processing to run in accordance with the reference trajectory,resulting in reduced machining accuracy.Ball screw pair is a transmission element that converts rotary motion into linear motion.Because of its high efficiency and high precision,it is often used in mechanical transmission systems.Therefore,the position error generated in the ball screw transmission is taken as the control plant in this research,and the error compensation technology using the golden section adaptive control and long short-term memory neural network based on the data-driven method is lucubrated respectively.The main research work of this paper is as follows:(1)The golden section adaptive control theory based on characteristic model is applied to the transmission error feedback compensation of the mechanical systemThe characteristic model-based golden section adaptive control method is introduced.The input and output data of the mechanical transmission system are used to build the characteristic model.Aiming at the established characteristic model of mechanical transmission system,the projected gradient method is used to identify the characteristic parameters online.The stability of the closedloop system composed of the characteristic model is analyzed.A simulation model of transmission error feedback compensation based on golden section adaptive control is built in MATLAB/Simulink.Through simulation analysis,the feasibility,effectiveness and anti-interference ability of this method for transmission error compensation are verified.(2)A long short-term memory neural network method using Bayesian optimization is proposed to predict and compensate the transmission error.Long short-term memory neural network is a deep model,and Bayesian optimization is used to optimize the hyperparameters of the model.Firstly,the error prediction model is trained by using the input and output data of the mechanical transmission system.Then,the trained neural network prediction model predicts the dynamic error according to the reference trajectory of the transmission system,and generates a new reference trajectory through pre-compensation,so as to ultimately reduce the error.By comparing the prediction error with the actual error,the prediction ability of the method is evaluated.Through the comparison of the error compensation effect between the proposed error precompensation method and the error feedback method with compensation mechanism,the feasibility,generality and superiority of this method are verified.(3)The application of adaptive golden section method based on characteristic model for error compensation control of the mechanical transmission systemOn the basis of MATLAB/Simulink simulation verification,the application of the characteristic model-based golden section adaptive control theory on the mechanical transmission platform is verified experimentally.Firstly,the mechanical transmission platform is introduced from the aspects of hardware,software and control circuit,and the structure and principle of error compensation are analyzed.Then,through experiments under various working conditions,and compared with the PID control algorithm,the feasibility,effectiveness and superiority of the golden section adaptive control are verified.After the local control is completed,a networked control platform for transmission error based on the iNetCon104 system is built,and the application of network control in transmission error compensation is explored.