Investigations On Cutting Parameter Optimization And Dimensional Error Monitoring System For Slender Bar Turning Operation

Slender bar plays a crucial part in mechanical industry. It is also well known thatslender bar turning operation is much more difficult than ordinary shafts turning oper-ation due to its low stiffness. At present, the slender bar turning is performed basicallyby empirical workmanship of the operator. This requires great skill of operator, andmakes the productivity of slender bar turning low. During slender bar turning pro-cess, the de?ection and vibration of the machined part are the main reasons affectingthe product quality, while these factors can be ignored in the non-slender bar turningprocess. Thus the approaches about the turning of non-slender bar are not well suitedfor the case of slender bar. To optimize the slender bar turning operations, it is verynecessary to investigate quantitatively dimensional errors and vibration during slenderbar turning operations.Based on the analysis of current research status of slender bar turning technol-ogy, a systematic research about dimensional errors and regenerative chatter of slenderbar turning is carried out in this work. Considering the fact that movable support isstill one of the most important accessories in slender bar turning operations today,special attention is paid on the effect of the movable support on dimensional errorsand cutting stability during machining process.Firstly, a numerical model is developed to predict dimensional errors in slenderbar turning with movable support. The proposed model includes the effect of themachine-tool-part-movable support elastic de?ections on dimensional errors.Usingthe predictive model, numerical simulations are provided to investigate the effects ofmounting method, movable support stiffness and cutting parameters on dimensionalerrors. Experimental results show that the developed model has the ability to predictefficiently the dimensional errors for slender bar turning operations.Secondly, stability analysis of regenerative chatter in slender bar turning withmovable support is performed to determine the critical chip width. Then, the effect ofmounting method, movable support stiffness, chip overlapping factor and machiningsystem dynamic parameters on critical chip width is studied by numerical simulations.Based on the aforementioned stability analysis, the measures are given to improve the chatter stability in slender bar turning process. Machining tests confirm the validityof the proposed cutting stability model.Thirdly, cutting parameters in multi-pass turning of slender bar with movablesupport are optimized in order to maximize production rate using genetic algorithm.Due to the low depth of cutting, multi-pass machining is generally needed in slenderbar turning operations. Thus, the objective function of multi-pass turning is built tominimize the average production time. Further, based on the obtained dimensional er-rors predicting model and the stability model of slender bar turning, the optimizationof cutting parameters is constrained to keep from regenerative chatter and satisfy thespecified product quality requirements. Experimental validations show that the pro-posed optimization method is both effective and efficient for slender bar turning withmovable support.Finally, artificial neural network is employed to develop an in-process moni-toring system of dimensional errors during slender bar turning process. The inputparameter selection method, combining the neural network modeling technique andorthogonal test, is designed and implemented to determine the optimal input parame-ter of the in-process monitoring model.The average effect of each candidate machin-ing factor and sensed information on the modeling performance is determined by theproposed method. The results show that the feed, the slenderness, the radial cuttingforce component, the axial cutting force component and the cutting distance providethe best combination to predict dimensional errors during slender bar turning process.Experiments show that the developed system has the ability to monitor accuratelydimensional errors within the range that they have been trained.