Research On Reliability Modeling And Assessment Of Critical Subsystems Of CNC Machine Tools Under Multi-source Information

The CNC machine tools are complex electromechanical systems.As the basic production equipment in manufacturing industry,it has been widely used in many business sectors,such as aerospace,rail transit,navigation and military companies.The failure of CNC machine tools may result in malfunctioning equipment of these engineering systems and lead to economic loss.Therefore,reliability assessment of CNC machine tools is an important requirement of industry and academia.With the“High-end CNC Machine Tools and Basic Manufacturing Equipment” major science and technology project and other major national science and technology projects,the technical level of domestic CNC machine tools has been significantly improved.In contrast,the reliability of domestic CNC machine tools is still not satisfactory,reliability has become a key issue restricting the development of domestic CNC machine tools.As typical complex electromechanical system,the reliability of CNC machine tools is determined by its critical subsystems.The reliability modeling and assessment of CNC machine tools critical subsystems is an important part of the reliability assessment of CNC machine tools.It is also important reference for machine tool design improvement,production plan optimization,and maintenance planning.The critical subsystems of CNC machine tools are different from ordinary mechanical products,which often show the characteristics of complex system structure,long development cycle,high manufacturing cost,complicated working condition stress and lack of reliability information.Due to the particularity of the work environment and task requirements,the characteristics of small batch customization are presented.This paper focuses on the reliability modeling and assessment of critical subsystems of CNC machine tools,the main research contents are summarized as follows:(1)A failure time data based reliability assessment method considering multi-source data and complex working conditions is proposed.For the failure time data of the CNC system can be obtained through the accelerated life test of the Original Equipment Manufacturers(OEMs)and the work site of user plant.This paper systematically studies the failure time data fusion modeling method combining multiple sources.By introducing the calibration factor into the acceleration model,a modeling method for merging multisource failure time data and characterizing the complex working conditions is constructed.Then the parameter estimation and reliability assessment methods based on Bayesian method are studied.(2)For the long-life and high-reliability products such as the CNC machine tool's spindle system,traditional reliability analysis methods based on failure-time data are inaccurate due to the insufficient analysis data.To address this problem,a performance degradation process method based on degradation data is proposed in this paper.For characterizing the individual heterogeneity among the product population,the random effect is introduced into the stochastic process models.The random effect leads to that the model parameter obeys a certain probability distribution,and the model parameter corresponding to each individual is a sample realization of the probability distribution.On this basis,the basic framework of parameter estimation of degradation model and reliability assessment of CNC machine tool's spindle system based on Bayesian method is studied.(3)A novel Bayesian fusion method is introduced to handle the unsolved challenges of degradation analysis with individual heterogeneity,small sample size and representation of complex conditions environmental.The Gamma process integrated random effects is used to describe the individual heterogeneity of degradation process.To characterize the complex working environment faced by the CNC machine tool's spindle system,the calibration factor is introduced into the degradation model,combined with the scale parameters and shape parameters.On this basis,a multi-source degradation data fusion modeling,performance evolution prediction and reliability assessment method based on Bayesian method is proposed.The fusion analysis of multi-source degradation data and the dynamic update of reliability assessment results are realized.(4)Bernoulli data,failure time data and the degradation data are three types of data commonly used in reliability assessment.All types of data can be collected from different sources for reliability assessment of CNC machine tool's spindle system.However,CNC machine tool's spindle system encounter the challenge of small sample size problems,i.e.,single-type data with limited sample size is insufficient to meet the requirement of highaccurate reliability analysis.Bayesian models are proposed to describe the inherent relationship between the Bernoulli data,failure time data and the degradation data,and further to integrate these three types of data for improving the accuracy of reliability analysis under small sample size situation.The deviance information criterion is used to select the appropriate model from the IG process,Gamma process and the Wiener process to model the degradation data.Reliability analysis is made based on the posterior distribution of the fusion model parameters with the aid of MCMC method and zerosones trick.