Establishment And Application Of Programming Loading Spectrum For Machine Tools Spindle Reliability Test Based On The Lamikiz Model

The reliability of machine tools is a critical symbol reflecting the development level of China's manufacturing industry.Since the spindle is a key functional unit of a machine tool,its reliability will affect the stability of the machining tool directly.Reliability test is an important means to improve the spindle reliability.At present,although the spindle reliability test equipment capable of simulating the real working condition has been independently developed,the dynamic program loading spectrum for reliability testing has not been established,which has become the main factor that constrains the rapid development of the spindle reliability bench test technology.Thus,the program loading spectrum establishment method for spindle reliability test has been systematically researched in this dissertation.And a series of problems,such as dynamic cutting force acquisition,time-domain extrapolation,load cycle counting method identification and the program loading spectrum classification,have been intensively studied.To take a step further,the spindle reliability bench test based on the established program the loading spectrum was carried out.The main contents of this thesis are written as follows:?1?The dynamic cutting force is the basic data for compiling the spindle program loading spectrum.Thus it is important to efficiently and accurately acquire dynamic cutting forces in the course of machining processes.The calculation process of dynamic cutting force is analyzed and summarized based on the Lamikiz model.It revealed that transient uncut chip thickness is one of the key parameters whose accuracy should be enhanced advisably.Thus,the assumption in the classical model where milling trajectory is considered as circular motion is optimized and improved in this dissertation,and an integral algorithm in polar coordinate system measuring the transient uncut chip thickness is proposed.Both the trochoidal motion of the tool point trajectory and the continuity of milling process are considered to be closer to the actual process conditions.Based on this,the calculated dynamic cutting force values obtained based on the improved model is compared with the test results.It proved that the calculation results have a good agreement with the test results.The relative errors of the cutting force in x and y directions are both close to 10%.However,in the cutting in and cutting out phase,the accuracy of the z-direction cutting force which is greatly affected by the impact load is not as good as the other two directions.?2?The dynamic cutting force coefficients,K_ic?i=t,r,a?and K_ie?i=t,r,a?are the other two important parameters directly affecting the accuracy of the Lamikiz model.Since the existing identification method has the problems of low solution efficiency,expensive test equipment and low versatility,a BP neural-network model with two hidden layers of 24 nodes is established in this dissertation.The model is used for cutting force coefficients prediction under different working conditions.A?five factors and four levels?average cutting force coefficient identification test is carried out based on the orthogonal experimental design principle.Five different factors,which are speed of rotation,depth of cut,cutter diameter,cutter teeth number and materials of work-piece,are chosen as characteristics of the training samples according to the test identification results.The identification results show that the accuracy rate of the prediction results generated by the proposed BP neural network model reaches as high as 80%.Thus,the model can be used to predict and identify the dynamic cutting force coefficients under different working conditions.?3?In order to fully obtain the dynamic cutting forces of the spindle service process and establish a more representative program loading spectrum,the Autoregressive Moving Average?abbreviated as ARMA?p,q??is adapted to fit and time domain extrapolate of the acquired spindle dynamic cutting force in this paper.The cutting forces with spindle speed median value is 595r/min are taken as an example.The extrapolation results obtained from it are compared with that from Hidden Markov model.It turns out that Mean absolute percentage error with the ARMA?p,q?model is much lower.And the first-order difference results of the cutting force in time domain has better stabilities.Both the autocorrelation function and the partial autocorrelation function results have obvious tailing characteristics.The RMSE value of axial and radial extrapolation results are respectively 72.48 and14.93,which are acceptable.Therefore,a large amount of effective cutting forces can be extrapolated with the ARMA?p,q?model and used for spindle program loading spectrum establishment.?4?Reliability program loading spectrum is a concentrated expression of load average,amplitude and frequency.It was established according to the former research results in this thesis.A 5-level rotational speed spectrum is established from 106 sets of on-site milling process parameters and its cutting forces in time domain.The percentage of each level has been determined by the method of kernel density estimation.They are respectively 52.0%,29.0%,8.0%,6.0%and 5.0%.Considering the effect of actual cutting force sequence strictly,a load cycle counting method based on damage mechanics theory is adopted in this thesis and the load-frequency results are averaged,amplitude solved and graded.On the basis of these results,program loading spectrum for spindle reliability bench test was established.Feasible technical solutions for the establishment of spindle reliability program loading spectrum have been proposed in this dissertation.Specific methods are provided for important segments and core steps.The program loading spectrum established in this dissertation provides theoretical and technical support for the spindle reliability bench test in the laboratory environment,and the development and improvement of the spindle reliability test specifications.