Thermal Error Modeling And Experimental Research Of The Spindle In NC Milling Machine

The main content of this dissertation is Thermal Error Modeling of the Spindle inNC Milling Machine. On the basis of fully analysis and sμmmarization of the studystatus on thermal errors modeling technique， the identification technique of thermal keypoints and the modeling methods of thermal errors are proposed. These models are allverified by experiment application from NC Milling Machine. The main research worksare shown as follows:（1）The experiment of thermal errors on NC Milling Machine. Using the SMTtemperature measuring device and laser CCD displacement sensors to measuretemperature and thermal error of the spindle in KXK-713NC Milling Machine.Thescheme is proposed and the data of this experiment are obtained.（2）the identification of thermal key points. According to the data series of thetemperature and thermal error measured on the spot, grey correlation model is used toanalyze the importance of each influencing factors in the temperature field of machinetools and gives performance appraisal of these factors and relative arrangement of theperformance. From this, the most key points of temperature sensors are selected, thedisposal of temperature sensors is optimize, the structure of thermal error modeling issimplified and the robustness is increased.（3）thermal error modeling and prediction based on neural network. The basicprinciple and method of BP and RBF neural network in thermal error modeling of thespindle in NC milling machine have been presented. Taking KXK-713NC millingmachine as the test object, the models based on BP and RBF neural network wereestablished and the prediction ability of these models were proved by experiments.（4）thermal error modeling and prediction based on grey system. The basic principleand method of grey model GM(1,N) in thermal error modeling of the spindle in NCmilling machine have been presented. Taking KXK-713NC milling machine as the testobject, the models based on BP and RBF neural network were established and theprediction ability of these models were proved by experiments.（5）composite prediction of thermal error on the basis of neural network and greysystem. Analyzing the advantages of neural network and grey system model, grey neuralnetwork model is established on the basis of BP neural network and Grey Model(1，N)．Through the prediction study on thermal error modeling in KXK-713NC milling machine the results are tested.By comparison of the predictable accuracy of these three models validated thesuperiority of grey neural network. This composite model not only has the neuralnetwork’s good ability in approaching nonlinear function but also has the accuracy OfGrey Model in making a prediction of small sample data and convenient modelingprocess. So the composite model has the higher predictable precision than these two alonemodels.