Remaining Useful Life Prediction Of Critical Components Of CNC Machine Tool Based On Weibull Regression Model

CNC machine tools are the basic equipment for the development of emerging technology industries and cutting-edge industries.Life prediction is the basis and key of health management and maintenance.This paper studies the remaining useful life prediction method based on operating condition information for CNC machine tools,minimizes the impact and loss caused by machine tool faults,and improves the service life and availability of machine tools.On the basis of fault correlation analysis,the multi-step diffusion of faults and the fault probability of components are considered,and the fault propagation diffusion coefficient is calculated based on the hypergraph theory,and a fault propagation diffusion model of CNC machine tools is constructed to describe the fault propagation mechanism.Combined with the model topology attributes,a comprehensive fault risk model is established,and the path fault risk is calculated to identify critical components.On this basis,the operation information collection,noise reduction,feature extraction and dimension reduction of critical components are processed.Weibull Regression Model is introduced to predict the remaining useful life of critical components under different operating conditions.The main research work of this thesis is as follows.(1)To make up for the defects of fault propagation model caused by ignoring the multi-step propagation of faults in the existing research on fault propagation of CNC machine tools,a fault propagation modeling method based on hypergraph theory is proposed.According to the historical fault information of CNC machine tools,the correlation analysis is carried out.In order to reduce the complexity of fault propagation research,a fuzzy clustering analysis method is introduced to perform module clustering on component units,and a hierarchical topology model of CNC machine tools is constructed.The Johnson method is used to correct the rank of fault data,and the fault probability models of each component and the whole machine considering the correlation of fault time are established.The hypergraph theory is introduced to synthesize the attribute index of the topology structure model to calculate the propagation diffusion coefficient to construct the fault propagation and diffusion model of CNC machine tools,and to describe the fault propagation and diffusion mechanism.(2)In order to solve the problem of identification deviation caused by identifying critical components based on a single indicator such as fault risk,combined with the component fault propagation mechanism,a critical component identification method fused with fault propagation diffusion model is proposed.The centrality measure of network theory is used to describe the topology attributes of component units,and the influence model of component unit fault propagation is constructed considering the differences of component unit fault propagation effects within and between classes.According to the fault mode frequency ratio and fault rate of component unit,the influence degree of fault mode and inherent fault risk of component unit are calculated by Analytic Network Process(ANP).The fault propagation influence degree of the component unit is integrated to calculate the component unit comprehensive fault risk and path fault risk,and identify the critical component units.(3)In order to obtain the operating condition information of the critical components of the CNC machine tool,by analyzing their structural characteristics,selecting the type of component condition monitoring data,and combining the force hammer test to determine the best monitoring point,building a component condition monitoring platform,and obtaining component condition monitoring data.Based on the obtained monitoring signals,trend elimination,noise reduction and feature extraction are carried out.Considering that different features have different responses to the degradation or fault of critical components of the machine tool,Principal Component Analysis(PCA)is applied to reduce the dimensionality of the signal features,remove the features with low correlation,and provide input for the construction of the remaining useful life prediction model of critical components.(4)In order to improve the traditional remaining useful prediction method based on a single working condition,which is cumbersome and time-consuming,a remaining useful life prediction method for critical components based on Weibull Regression Model(WRM)is proposed.According to the condition monitoring information of critical components of CNC machine tool,considering the operating conditions of machine tool,taking the new characteristics after dimension reduction of PCA as internal covariates and operating load and speed as external covariates,a Weibull regression model fully considering the operating condition information of machine tool is established,based on which the remaining useful life prediction of critical components of machine tool is carried out.The performance of the built model is evaluated by various model performance evaluation methods,which verifies the rationality of the proposed method and the necessity of considering the external working conditions to predict the remaining useful life of the critical components of the machine tool.