Research For The Spindle Sensor Placement Optimization And Recovery Of Fault Data

In the industrial manufacturing industry CNC machine has been developing towardsHigh-speed and precision direction and thermal error has become one of the main factorsinfluencing the machining accuracy of machine tool. According to the problems of thetemperature sensor placement optimization technical difficulties and being lack of studyabout thermal error compensational system stability, the thesis, which regards theprocessing center of VMC750E and VDF850D as the researching object and thecompensation method of thermal error to improve machine processing precision as theresearching aim, systematically research the key techniques of thermal error compensationsuch as placement optimization of the temperature sensor, modeling thermal error andstabilizing the thermal error compensation system. The main contents are:By doing the experiment on the temperature characteristics using five-point methodin VMC750E machining center, and analyzing characteristics about spindle in differentspeed between the temperature sensor temperature data (heat source point, Non-sourcepoint), between the heat source point temperature data and spindle axial thermaldeformation, it is concluded that the characteristics of the non-linear, coupling betweenthem.Research for the temperature sensor placement optimization method based on thethermal deformation decomposition theory. By doing the experiment on heating-coolingprocess temperature on variable condition in the machine center VDF850D, draw theconclusion by analyzing that machine column thermal deformation should not beneglected in thermal error of machine tool spindle axial, and spindle axial thermaldeformation include main spindle thermal deformation, and machine column thermaldeformation. On this basis, use cluster analysis to achieve temperature sensor placementoptimization of main spindle.According to the nonlinear characteristics and the characteristics of limitation of datasample between the machine tool spindle thermal error data and the machine tooltemperature sensor measuring temperature data, by using the combination kernel functionleast squares support vector machines (SVM) algorithm establish spindle thermal errorprediction model, it shows the advantages of a small sample, adaptable, shorter trainingtime, the generalization ability. Embedded in HNC-21system and make experiment ofmachine tool thermal error real-time compensation, and thermal error reduces to less than10μm by35μm.At last based on the characteristics of information entropy that can measure nonlinearrelation between data and quantify data uncertainty, the thesis puts forward quantifying temperature data uncertainty of measurement methods on the variable working conditionand analyzing the characteristics of the data of temperature by using quantitative measureof hypothesis test to diagnosis temperature data anomalies, through the LS-SVMregression estimation repair abnormal temperature data ensuring stability of thermal errorcompensation system.