Research On The Thermal Errors Compensation Techniques Of High Speed CNC Machining Center Motorized Spindle

High speed and high precision of the machine tools have become a uppermost development direction of the current and future’s manufacturing equipment. With the machine tools’manufacturing and assembly accuracy rising ceaselessly, the mechanical errors of machine tools is going down in the proportion of the total errors, at the same time, the error caused by the heat has become the main error source of precision machine tools, therefore, the research on thermal errors compensation is a research direction with high value. In this thesis, the milling motorized spindle of CX8075turning and milling combined CNC machining center has been studied in detail, and the following basic research work about compensating its thermal errors has been carried out.Firstly, based on the heat transferring and finite element analysis theory, by using finite element analysis software ABAQUS, the finite element model of CX8075milling motorized spindle is established,and the analysis for its thermal-structural characteristics is done. The steady temperature field distribution, the data of temperature and the data of thermal errors from the beginning to reach the thermal equilibrium state are obtained through the analysis.Secondly, based on the results data of finite element analysis and the theory of fuzzy clustering, the clustering analysis for the temperature variables is done. And then, some typical temperature variables are chosen due to the correlation between variables. Two sets of temperature of variables and their combinations are chosen for modeling the thermal errors compensation models of the motorized spindle at axial and radial directions after the combinatorial optimization for the typical temperature variables.Finally, two kinds of thermal errors compensation models are established respectively based on multiple linear regression and BP neural network. By comparing the fitting effects and prediction errors of the two different models, it is found that the thermal errors compensation model based on BP neural network has a higher fitting and forecast precision, its robustness is better also, and it is suitable for the thermal errors compensation of the motorized spindle.