Research On Selection Of Robust Temperature-Sensitive Points And Thermal Error Modeling Of Temperature Fields For Machine Tools

Since the reform and opening up,China’s machine tool industry has gradually developed and grown,with the continuous development of industrial technology,the requirements of machine tool accuracy is also increasing,high precision CNC machine tools can largely reflect the ability of the national manufacturing industry.The thermal error of CNC machine tools is the largest source of error affecting the machining accuracy of machine tools,and it is important to study the reduction of thermal error to improve the machining accuracy of CNC machine tools.The analysis,modelling and prediction of thermal errors are divided into physically based methods and data-driven methods.The actual machining conditions,cutting process parameters,coolant use and ambient temperature and other complex error sources have an impact on the thermal errors of machine tools.Therefore,the data-driven approach has attracted increasing attention,and many scholars both at home and abroad have conducted extensive research to explore the internal characteristics of process data and to establish the mapping between temperature and thermal errors.This paper takes the VMC1165 B vertical machining centre as the research object,conducts long-term extensive experiments and data analysis,and combines the existing modelling theoretical basis to investigate the temperature sensitivity point selection and thermal error modelling for the temperature field robustness of CNC machine tools,the main research content is as follows:1)Design the experimental scheme of thermal characteristics of CNC machine tools according to ISO230-3 international standard,design the software and hardware of the testing system and debug the testing system;collect temperature and displacement data through Labview software platform,realise the sub-window and total window display and data saving of temperature data as well as the real-time display and saving of thermal deformation data,and filter the measurement data by Butterworth digital filter of the software Thus reducing the influence of external high frequency interference on the thermal error measurement in the actual measurement process.2)Measurement and analysis of the temperature field and thermal deformation.The temperature field and thermal deformation generated by the spindle rotation,the temperature field and thermal deformation generated by the x-axis and y-axis linear axis feed motion are measured and analysed,and then the temperature field and thermal deformation under the whole machine motion with different spindle speed and linear axis combined speed in x and y directions are measured and analysed.3)The temperature data and displacement data are collated and data enhanced to produce a data set for robust temperature sensitivity point selection.Fuzzy clustering combined with Pearson correlation coefficient method was used to select the robust temperature sensitive points and compared with the non-robust temperature sensitive points selected by fuzzy clustering combined with grey correlation degree for verification.4)Machine tool thermal error modelling.Based on the random forest(RF)algorithm for thermal error modelling,the regression tree part of the model is visualised to make the model interpretable and compared with the multiple linear regression(MLR)model,BP neural network model and partial least squares(PLS)algorithm thermal error model for analysis and verification of their prediction performance.