Research On Key Technologies For Compiling Load Spectrum Of CNC Lathe Based On Kernel Function

The load spectrum is the basic data to evaluate the product life and to guide the reliability test and reliability design of the product.In this paper,the kernel density estimation method is applied to the load spectrum compiling process in the field of CNC machine tools,and the key technologies such as determination of typical test conditions,load measurement and signal processing analysis,load extrapolation and program load spectrum compiling in the process of load spectrum compiling are studied.The main contents are as follows:(1)Based on kernel density estimation principle,in terms of kernel function selection,based on the physical significance,asymmetry and generalization characteristics of each parameter of weibull distribution,this paper proposes a kernel density estimation method based on weibull kernel function,which is used to estimate the probability density of one-dimensional variable data in the process of loading spectrum compilation.The standard gaussian kernel,adaptive bandwidth and l-order direct insertion method were selected to estimate the probability density of two-dimensional variable data during the preparation of the load spectrum.(2)According to the load process data collected by users on site,and on the basis of statistical analysis of the machining information of each machining mode and each working condition,this paper proposes a method to determine the typical test working condition of CNC lathe based on kernel density estimation.In addition,a turning load test system was set up in the laboratory,and a test scheme was developed to carry out the dynamic turning load test.(3)The trend term,singular point and noise reduction of the load signal in time domain are removed.The load is transformed from time domain to rain basin by rain flow counting.The Markov Chain-Monte Carlo method is used to synthesize the cutting load data in multiple working conditions.(4)Based on the advantages of the non-parametric rain flow extrapolation method,which is not affected by the complexity of the rain flow matrix,is independent of the matrix distribution,and can extrapolate the load and cycle frequency at the same time.The two-dimensional kernel density estimation is carried out for the from-to matrix of the three-direction cutting load under different machining methods,and the probability density distribution of the initial and final value of the load is obtained.According To the probability density distribution,the long-term load cycle extrapolated To the specified frequency is obtained by Monte Carlo method,and the extrapolated from-to matrix of the three-direction cutting load of CNC lathe under different processing methods is obtained.(5)The extrapolated from-to matrix was transformed into mean-range matrix,and the two-dimensional kernel density estimation was carried out To obtain the probability density distribution of load-mean amplitude.Using Conover ratio coefficient of load is divided into eight interval,mean and amplitude based on two-dimensional mean-amplitude probability density function and total frequency extrapolation cycle,solving all the total cycles and amplitude of each interval,namely get all two-dimensional mean-amplitude loading spectrum,used to guide the subsequent numerical control lathe key features reliability test,etc.