| CNC machine tools are the foundation of the manufacturing industry,and their technical level determines the height of the development of the Chinese manufacturing industry.At present,there is a big gap between the market competitiveness of high-end machine tools in our country and foreign machine tools,one of the main reasons is that the reliability level of domestic machine tools is low.In-depth research on reliability allocation methods to ensure and improve the reliability level of domestic machine tools,and then enhance the market competitiveness of domestic machine tools,is of great significance to promote the overall development of the Chinese manufacturing industry.CNC machine tools belong to the complex equipment of machine,electric and hydraulic,which have the characteristics of complex structures and diverse functions.There is a lot of fuzzy information in the reliability design of CNC machine tools,and most of the existing reliability design methods make it difficult to express this fuzzy information completely and accurately.In addition,the existing reliability allocation research neglects the correlation between the influencing factors in reliability allocation and the difference in the actual reliability level of the machine tool under different working conditions,which leads to the lack of accuracy and rationality of reliability allocation results.With the advent of the era of big data,the relevant data that can be used for the reliability design of CNC machine tools have been rich and accumulated,but these historical reliability data cannot play a role in the existing reliability allocation research.In view of the above problems,this thesis takes domestic CNC machine tools as the research object,based on fuzzy mathematics theory,proportional hazards model and deep learning method,considering a variety of factors in the operation process of machine tools,combined with the actual working conditions of machine tools,carries out the research on the reliability allocation method of CNC machine tool.The main work and innovative achievements of this thesis are as follows:(1)The rough probabilistic linguistic term set that can fully express and process fuzzy information in the reliability design of CNC machine tools is proposed,and the traditional failure mode and effects analysis(FMEA)method is improved.To solve the problem that it is difficult for experts to express and process fuzzy information flexibly and completely in the current reliability design,this thesis proposes the rough probabilistic linguistic term set based on probabilistic linguistic term set and rough set theory,which not only maintains the flexibility of probabilistic linguistic term set but also enhances its ability to express and process fuzzy information in complex linguistic environment.On this basis,the FMEA method based on the rough probabilistic linguistic term set is proposed.Taking a domestic machining center as an example,FMEA is carried out based on the historical fault data of the machine tool,and the key subsystem of the machining center is determined,which lays a foundation for the subsequent reliability allocation of the machine tool and the correction of the reliability allocation results of the key subsystem considering the influence of working conditions.(2)A two-stage reliability allocation method is proposed,which comprehensively considers the weight and correlation of the influence factors.To solve the problem of ignoring the correlation among influencing factors and misusing the ordering information as proportional information in the reliability allocation process of CNC machine tools,the rough probabilistic linguistic Muirhead mean operator and the rough probabilistic linguistic best-worst method are proposed to realize the accurate processing of fuzzy evaluation information in the reliability allocation process of machine tools.On this basis,a two-stage reliability allocation method for CNC machine tools is proposed.In the first stage,the weight and correlation of the influence factors are comprehensively considered,and the fault rate ranking of the subsystem is calculated by using the rough probabilistic linguistic Muirhead mean operator;In the second stage,the reliability allocation weights of the subsystems are calculated using the rough probabilistic linguistic best-worst method.Taking a domestic machining center as an example,combined with the historical reliability evaluation results of the machine tool,the accuracy of the reliability allocation results of the machine tool is verified comprehensively through the combination of qualitative and quantitative methods.(3)Based on the proportional hazards model,the reliability allocation result correction method of the CNC machine tool is proposed.Aiming at the difference in the actual reliability level of machine tools under different working conditions,a proportional hazards model that can accurately represent the relationship between reliability allocation results and actual working conditions is established.Aiming at the problem of lack of reliability data in the early stage of machine tool design,the virtual fault data of the target machine tool subsystems is generated by random sampling method according to reliability allocation results of the machine tool and reliability evaluation results of existing similar subsystems,and the covariate coefficient of proportional risk model is accurately estimated.Taking a domestic machining center as an example,the proportional hazards model of the key subsystem of the machine tool is established,and the reliability allocation results of the machine tool are refined and revised.(4)Aiming at the problem that it is difficult to make full use of the historical reliability data of CNC machine tools in the reliability allocation process,a reliability allocation model of CNC machine tools is established which can make full use of historical reliability data.Based on the comprehensive analysis of the characteristics of historical reliability data of machine tools,the original reliability data set is established,and the conditional tabular generative adversarial network is used to expand the original data set.On this basis,the reliability allocation model of the CNC machine tool based on multi-source historical reliability data is established,and the initial parameters of the model are optimized by genetic algorithm to reduce the randomness of the initial parameters generated by the model.With the accumulation of reliability data of CNC machine tools,the accuracy of the model will continue to rise.Sufficient historical data will reduce the influence of subjective factors of experts and achieve more refined reliability allocation.Through the research of this thesis,the processing method of multivariate fuzzy information in the reliability allocation process of CNC machine tools is proposed,and the reliability allocation result is combined with the actual working condition,which improves the rationality and accuracy of the reliability allocation result of CNC machine tools,realizes the fine reliability allocation of CNC machine tools,which provides technical support for reliability design of CNC machine tools. |
PDF Full Text Request