Preparation Of Servo Tool Holder Power Head Status Monitoring System And Remaining Useful Life Prediction Based On Data Drive

As the ballast stone of industrial development,CNC machine tools have an important reference for measuring the level of a country's manufacturing industry.Domestic CNC machine tool manufacturing technology and international advanced technology are relatively different,and there is still a big gap in the level of reliability,especially in the manufacture of core components of CNC equipment.Among them,the servo tool holder,as the core component of the turning center or the turning center,has always relied on imports.Although China's knife holder manufacturers have obtained some technological progress with the help of national policy support in recent years,and initially have the ability to produce and manufacture core components of CNC equipment,their advanced functions cannot be maintained for a long time,and the reliability problem is becoming more and more prominent.Prognostics and Health Management(PHM)is an important technical means to effectively improve the reliability of product use,and can also specify maintenance strategies in advance,reduce maintenance costs,and reduce downtime losses.Remaining useful life prediction has always been a frontier hot spot in PHM research,especially in the early stage of system degradation,when the degradation that has occurred has not caused excessive losses,and the residual life(RUL)study of the system can grasp the health of the system and can also provide a reference for subsequent maintenance measures.In this paper,the remaining life of a domestic servo tool holder power head system is studied,and the main work content is as follows:1)The structure and fault analysis of the subject-servo tool holder power head system are carried out,and the sensor arrangement scheme is provided.First of all,the structural analysis of the domestic servo tool holder power head system is carried out,and the mechanical structure and transmission principle of the power head system are clarified.Secondly,the servo tool holder fault tree analysis(FTA)and FMECA analysis were carried out,and the structure and fault mechanism related to the top time were analyzed,and the three typical failure modes and main fault parts of the power head system were determined.Finally,according to the above analysis,the sensor arrangement during the test monitoring process is provided as a reference.2)Design and build a servo tool holder power head system condition monitoring test system,combined with the purpose and conditions of the test to develop a test plan.The system mainly includes the power unit of the subject object,the cutting load simulation loading device and the status signal monitoring and acquisition device.The power unit mainly includes the optional power servo motor and the adapted driver;the cutting load simulation loading device mainly includes the eddy current dynamometer,the drive instrument,the measurement and control instrument,etc.;the status signal monitoring and acquisition system mainly includes sensors,data acquisition cards and data monitoring and acquisition software based on Labview software development.According to the current reliability test and accelerated life test related theories and tests,the life prediction test scheme of this subject is formulated.3)Based on the nuclear principal element analysis method,the comprehensive health index of the power head system is established.Firstly,in order to select channels with more information in the channel,a sensor selection method based on information entropy is used;secondly,in order to avoid too much redundant information,new information is calculated for verification;finally,the comprehensive health index of the power head system is constructed based on the analysis of the main element of the kernel,and the indicator of good performance of the degradation trend is obtained.4)Compared the effects of two feature selection methods on the prediction accuracy of the remaining life of the power head(RUL),in addition to the health index extracted based on KPCA,the health index obtained after the general PCA fusion was used.After the exponential degradation model is introduced,a parameter update model based on Bayesian thought is established as a prior distribution,and the posterior distribution is solved by MCMC method.The final results show that the health index established after the acquisition of characteristics by using information entropy is more pure and the accuracy of life prediction is higher.