Study On Key Techniques For Predicting Contour Errors Of High-speed CNC Machine Tools

CNC machine tools are the "worker" of the equipment manufacturing industry,its performance directly reflects a country's production and manufacturing capability.Accuracy and efficiency are two important indicators to measure the performance of CNC machine tools,which are reflected in the contour accuracy and processing speed of the tool trajectory during the machining process of the machine tool.In high-speed machining,thermal errors and servo control system errors are the main factors affecting contour accuracy.Therefore,the high-precision predictive the contour error of the tool trajectory caused by servo control system error and thermal errors is an important basis for guaranteeing machining accuracy.In order to predict the contour error in the machining process with high precision,it is necessary to study the key technologies of servo control system error modeling,thermal error modeling and contour error estimation.Combined with the current research status at home and abroad,the main research content and innovative results of this thesis are as follows.(1)The main parameters and characteristics of the system are analyzed,and the modeling method of servo control system combined with principle analysis and convolutional neural network learning is studied.Based on the principle of generality,the principle of the main parameters of the system is analyzed,which simplifies the modeling process.A universal and easy-toimplement principle modeling method for the servo control system is proposed,which solves the problem of optimizing the model structure by considering more physical factors in the existing principle model,resulting in a complex modeling process and a lack of generality in the model.On this basis,through the analysis of the characteristics of the servo control system,a deep convolutional neural network for timing problems is established to learn other unknown complex relationships of the actual servo control that the principle model lacks.The above methods are used to improve the accuracy of the servo control system model,and solve the problem that the existing empirical modeling method based on machine learning lacks the ability to learn the characteristics of the servo control system.The accuracy of the model was verified through experiments and compared with existing research.The experimental results show that the model established in this thesis has better versatility and higher accuracy.(2)Based on the analysis of the characteristic information of high-quality samples,the operation trajectory of the servo control system is constructed,and a method for generating high-quality sample sets for system identification is proposed.The main factors affecting the tracking error of the servo control system are analyzed,the sample feature domain is defined,and the method of generating high-quality sample feature data is proposed based on the principle of low-disparity sequence sampling.The relationship between the sample characteristics and the experimental trajectory is analyzed,and the construction method of the experimental trajectory is studied to obtain input and output samples for model identification.The method of reverse construction of experimental trajectory based on sample characteristics proposed in this thesis can guarantee that the acquired sample set contains sufficient effective feature information,and solves the problem that existing research only selects experimental trajectories based on experience,lacks theoretical basis,and leads to insufficient feature information of sample sets,which affects the accuracy of the model.By comparing with the trajectory used in the existing literature,the samples acquired based on the trajectory constructed by the method in this thesis can effectively improve the identification accuracy of the model.(3)A data enhancement method for constructing multiple types of temperature field distributions through a small number of experimental samples is proposed,combined with the constructed ANSYS model,and a method for efficient modeling of neural network thermal errors is proposed.The relationship between the motion state of each axis of the machine tool and the temperature field distribution is analyzed,and a method for enhancing the temperature field distribution data based on small sample experimental data is proposed.A feature set containing a variety of temperature field distributions is constructed.It solves the problem that the existing thermal error modeling sample data research mainly focuses on the location and number of measurement points of the temperature field distribution,and lacks exploration of the diversity of the temperature field distribution,resulting in insufficient sample characteristics.Combined with ANSYS simulation instead of the experiment to obtain rich and diverse temperature distribution-thermal error sample data.The problem that the prediction accuracy of the neural network thermal error model is affected by the inability to obtain enough sample data through experiments is solved.The experimental results show that the samples obtained based on the sample enhancement method proposed in this thesis improve the prediction accuracy of the thermal error model,which verifies the effectiveness of the method.(4)A new method to estimate the contour error of the tool trajectory is proposed,and a general contour error model of the tool trajectory is established.The contour error of the tool trajectory without datum is defined.A non-datum contour error estimation model is established based on the principle of minimum area.The contour error with datum is regarded as a special case of the model,which solves the problem that the existing contour error estimation method is not applicable to all objects.When the tool orientation error needs to be considered,the tool tip position and tool direction information are integrated by introducing the effective cutting length of the tool,and a unified representation method for the tool tip position and tool orientation error is proposed,which can more accurately reflect the quality of the part.The problem that the existing research does not comprehensively consider the tool tip position and the tool direction is solved,which affects the contour error of the tool trajectory.The experiment verifies the validity of the non-datum contour error calculation method and the advantages of the comprehensive contour error model of the tool tip position and the direction of the tool...