Research On Prediction Of Thermal Error And Cutting Vibration In NC Machining Process Based On Operation Conditions

In high-speed precision machining, with the influences of mechanical characteristics,thermal characteristics and thermal-mechanical coupling characteristics of NC machine tools,thermal error and cutting vibration have become to be the key factors which prevent theimprovement of machining quality and efficiency. As complicated electromechanicalequipments, NC machine tools are considered to be typical nonlinear systems. In thisdissertation, based on the theory of nonlinear system identification and control, as well asHertzian contact theory, neural networks, finite differential theory and signal processing theory,the relationships between operation parameters (spindle speed, feed rate, and axial cuttingdepth) and thermal positioning error of feed system in NC machine tools are systematicallyinvestigated, so as the relationships between operation parameters and cutting vibration inmachining, and a control method to suppress thermal positioning error of feed system andcutting vibration by adjusting operation parameters is presented.According the fact of thermal contact resistance brought by the imperfect contact betweenrolling elements and raceway of the supporting bearings in feed system, a theoretical method tocalculate the thermal contact resistance is presented based on Hertzian contact theory andLaplace temperature distribution equation in ellipsoidal coordinates, and the internal loadingdistribution and elliptic contact area parameters are carefully studied. According the fact offriction heating brought by the contact friction between rolling elements and raceway of thesupporting bearings, a theoretical method to calculate the friction heating rate is presented withcareful consideration of the spinning and sliding friction torque in the elliptic contact area.Furthermore, the influences of operation parameters on the thermal contact resistance andfriction heating rate are analyzed, and experimental verifications were carried out on self-madequasi high-speed feed system.Based on the orthogonal experiments carried out on the self-made quasi high-speed feedsystem, the influences of operation parameters on the temperature-rises of heat sources in feedsystem are analyzed, and an operation rule database is built by defining operation distance andclustering analysis. Furthermore, on the basis of nonlinear system identification modelNARMA-L2and wavelet neural network, the relation model for operation parameters andtemperature-rises of heat sources in feed system is obtained using improved particle swarmoptimization (PSO) as the training algorithm, and the temperature-rises of heat sources can bepredicted with operation parameters as the inputs.With the temperatures, the thermal contact resistances and friction heating of supportingbearings and screw-nut as the boundary conditions, the heat conduction equation of ball screwin feed system is solved numerically using improved group explicit finite differential method, and the temperature and thermal deformation distribution of the ball screw under differentoperation parameters are obtained. Moreover, the thermal positioning error of the feed systemcan be predicted with operation parameters as the inputs by establishing the relationshipbetween the thermal deformation of the ball screw and thermal positioning error of the feedsystem. Based on contact analysis, the finite element parametric model is built using ANSYSparametric design and analysis language, and temperature and thermal deformation distributionof the feed system is solved.Finally, the results of numerical solution and simulation solutionare compared.With impeller milling process as an example, the vibration frequencies in milling areanalyzed based on the dynamic model of milling process. Using wavelet packet analysis, thechanges of frequency transition and harmonic amplitude of vibration signal in different cuttingstates, and the key features for detecting vibration sates are presented. Furthermore, on the basisof nonlinear system identification model NARMA-L2and wavelet neural network, the relationmodel for operation parameters and cutting vibration in milling is obtained, and the cuttingvibration can be predicted with operation parameters as the inputs.Using Multi-class Least Squares Support Vector Machines, the thermal positioning errorof the feed system is identified with the mean, variance, mean square value and maximum ofthe temperatures of supporting bearings and screw-nut as feature vector, and cutting vibrationstates in milling are detected with the relative wavelet packet energy and standard deviation ofthe multiple-frequency vibration as feature vector. Based on the prediction of thermalpositioning error of the feed system and cutting vibration, the adjusting strategies of operationparameters are presented according to the Pareto sets and Pareto Fronts of a multi-objectiveoptimization, and a experimental verification was carried out in impeller milling to suppress thethermal positioning error of feed system and the cutting vibration by adjusting operationparameters.The achievements of this dissertation provide an effective approach to detect and controlthe thermal positioning error of feed system and the cutting vibration in NC machining process,which is proved to have both theoretical significance and practical application value.