Research On Contour Error Estimation And Control Method Of Two Axis Feed Servo System

With the continuous development of manufacturing technology,more and more manufacturing processes involve high-precision machining,and precision motion systems such as CNC machine tools have been widely used in high-end manufacturing industries.However,with the increase of feed speed,the inherent frequency of the mechanical transmission system limits the response speed of the feed system,and the mismatched kinematic relationship between different axes leads to contour errors in multi-axis linkage machining.Therefore,research on contour motion control for multi-axis precision motion systems has important theoretical significance and significant engineering application value.In response to the problems of inaccurate contour error estimation and poor timeliness in motion control of feed servo systems,inaccurate prediction of future contour errors in contour error prediction control,and the inability of contour error analysis models to provide accurate basis for trajectory pre-compensation,this paper specifically focuses on the following research contents:Firstly,addressing the issue of estimation distortion in existing contour error estimation methods,a contour error offline estimation method is proposed.This method transforms the problem of contour error solving into an extremum solving problem and uses the Newton method to solve it.Additionally,aiming at the poor real-time performance of existing contour error real-time estimation methods,a contour error online estimation method is proposed.The method simplifies the Newton-based contour error offline estimation method using the quasi-Newton method and accelerates its iterative process through the Steffensen’s method to reduce the time for real-time estimation of contour error.The accuracy of both offline and online estimation methods was verified through comparative experiments,and the real-time effectiveness of the online estimation method was also validated.These methods lay a foundation for subsequent contour error offline and online control.Secondly,a contour error prediction method is proposed to address the issue of inaccurate prediction of future contour errors in contour error prediction and control.The method utilizes an LSTM neural network to establish a contour error prediction model and uses the POA algorithm to optimize the hyperparameters of the prediction model.By using randomly generated NURBS curves to construct the training dataset and through reasonable training feature selection,the training of the contour error prediction model is completed.Experimental results show that by selecting appropriate training features,the proposed contour error prediction model has higher prediction accuracy.Furthermore,compared with other neural network models in comparative experiments,the proposed method performs better in predicting contour errors and has better performance in contour error prediction.Then,in response to the problem that the contour error analytical model cannot provide accurate basis for trajectory pre-compensation,a contour error compensation method is proposed.To achieve accurate compensation for contour error,the Simulink/Simscape multi-domain modeling tool is used to separately model the electrical subsystem,control subsystem,and mechanical and system subsystems of the feed system.To achieve accurate decoupling of contour error,a transformation task coordinate system is established based on the Frenet coordinate system principle,and the computed compensation components for each axis are directly compensated to the desired trajectory to achieve compensation for contour error.Finally,through the established two-axis feed servo experimental platform based on Ether CAT communication,the offline control,online control,and predictive control of contour error proposed in this paper were experimentally verified.The effectiveness of the offline control and online control proposed in this paper for contour error control was verified by comparing them with PID control and cross-coupling control.The effectiveness of the predictive control proposed in this paper was verified by comparing the contour error before and after prediction compensation.