| Summary: | 博士 === 大同工學院 === 機械工程研究所 === 86 === Theraml Analysis of Creep Feed Grinding The present study focus on the thermal analysis of creep feed grindingwhich includes the experimental measurement and theoretical analysis. In the beginning, a thermal model based on the heat transfer behaviours among the fluid, workpiece and grain is developed. It is found that the conduction behaviourin the workpiece moving direction is significant and innegligible for creep feed grinding. Therefore, this behaviour is considered in this model so as to give a whole picture of creep feed grinding. Moreover, the thermal distribution amongthe workpiece, fluid and wheel can be realized from the defined thermal partitionratios. Thus, the cooling effects of coolant are shown to be more crucial especially at the characteristic working conditions of creep feed grinding, i. e., at a large grinding depth and a small workpiece speed. Next, sequences of experimental measurement are performed. The predicted workpiece emperatures show great agreement with experimental and published data. Thus, the validity of the thermal model is proved. The critical grinding energy for nuclear film boiling is derived due to the present thermal model. Itis found experimentally that the grinding burning occurs through a transitionprocess. As the nuclear film boiling of the fluid occurs, the grinding energy obtained experimentally has been found to be greater than the critical grinding energy. Then, the cooling effect of the fluid continuously becomes worse. Finally, workpiece burning happens. Therefore, once the grinding energy is greater than the critical grinding energy, a great of possibility of the workpiece burning may happen. Subsequentally, the grinding force of creep feed grinding are modeled and forecasted by using back propagation neural (BPN) network. The BPN networkis improved by integrating an error distribution function (EDF), a process that is proved to be useful in overcoming local minimum problems effectively and in accelerating greatly the convergence speed. Thus, the implemented neural networkalgorithm can be used to predict the grinding forces. Furthermore, the prediction of the grinding energy served as heat source in grinding process can be obtained due to the grinding force. Thus, the predicted grinding energy can be applied to the thermal model in forecasting the workpiece temperature.Cooperated with the critical grinding energy obtained from the presented thermal model, the predicted grinding energy can also forecast the workpiece burning so that burning can be prevented. Finally, a selection scheme of working conditions in view of the avoidanceof the workpiece burning is presented. Considering the working efficiency, the working conditions are selected to maximum metal removed rate, MRRmax. The results suggested that a lower wheel speed as well as a larger size of wheel isavailable to have a better working effeciency.
|
|---|
