Research On Thermal Error Modeling Of Motorized Spindle Based On Thermal Deformation Mechanism And Data Driven

With the advent of the Industry 4.0 era,advanced manufacturing technology has developed rapidly.As an aircraft carrier in machining,heavy duty CNC machine tools have put forward higher requirements for its machining accuracy.Thermal error is an important source of error for heavy duty CNC machine tools.As the core component of CNC machine tools,thermal deformation of the spindle caused by the heat generated by motor loss and bearing friction is one of the main factors affecting machining accuracy.Modeling and compensating the thermal error of the spindle is a key step to improve the machining accuracy of the machine tool.Therefore,how to build an efficient and accurate thermal error prediction model is a big problem to be solved in the field of thermal error research.In this paper,KSB-225 motorized spindle of the heavy-duty gantry drilling machine is the main research object.Based on the analysis of the heat transfer theory of the motorized spindle,a temperature field model of the spindle was constructed.Based on this model,a generalized constrained neural network thermal error modeling method based on mechanism and data driven fusion is proposed.The main research work of this paper is described as follows:(1)Analysis of thermal characteristics of motorized spindle and construction of temperature field model.First,the thermal deformation mechanism of the motorized spindle is analyzed to determine the main influencing factors that affect the thermal deformation of the spindle.Then,a finite element model of the motorized spindle is established,and the thermal characteristics are analyzed to determine the relationship between the heat source,temperature,and thermal error.Finally,based on the thermal modal theory and heat source method,this paper analyzes and calculates the transfer process of the motorized spindle,constructs a temperature field mechanism model,points out the temperature change rule.It provides a theoretical basis for the study of the thermal characteristics of the heat source and the prediction of the temperature field.(2)The method of parameter identification and correction of temperature field model based on Kalman filter.First,by analyzing the structure of the spindle and the internal and external heat sources,this paper establishes a finite element analysis model of the spindle box,determines the position of the temperature measurement point,and builds a data monitoring system.Then,aiming at the difficulty of determining the parameters of the temperature field model,considering the strong applicability of the nonlinear Kalman Filter method in the identification of nonlinear systems,a parameter identification method based on unscented Kalman filtering is proposed.By discretizing the temperature field model,the parameters are identified.Finally,considering the increase in the amount of data over time,there is a risk of filtering divergence,a parameter correction method based on process noise covariance matrix(Q)and measurement noise covariance matrix(R)adaptive untraced Kalman filter is proposed.Through the dynamic adjustment of the Q and R matrices,the prediction accuracy of the temperature field model is improved,which lays the foundation for the subsequent construction of a mechanism-based thermal error model.(3)Thermal error modeling method based on generalized constrained neural network.Aiming at the problems of low transparency and data dependence of the traditional data-driven thermal error model,considering the interpretability of the mechanism thermal error model,a generalized constraint neural network(GCNN)based motorized spindle thermal error modeling method was proposed.Based on the Bat Algorithm-Wavelet Neural Network(BA-WNN)thermal error model,this paper introduced Partially-known Relationship(PKR)based on mechanism thermal error model.By constraining the parameter set of the BA-WNN target,the mechanism model is integrated with the data-driven model,which enhances the transparency of the model while improving the prediction accuracy and robustness of the model,and provides a new idea for modeling the thermal error of the motorized spindle...