Research On Health Management For Critical Component Of Mechanical Equipment Based On EEMD-Mahalanobis Taguchi System

With the progress of science and technology and the rapid development of precision instruments,the reliability and safety requirements for mechanical equipment are also gradually increased.Without a comprehensive set of management systems to monitor and diagnose mechanical equipment,it is difficult to ensure effective,smooth and continuous operation.Once the critical components of mechanical equipment fail,it will result in a variety of burdens,reduce operational efficiency,and even lead to a series of adverse effects.Mechanical equipment inevitably undergoes the process from being normal to degenerate and finally fail in the operation phase.Therefore,only by managing the health of mechanical equipment,and obtaining real-time monitoring information to accurately and effectively perform fault diagnosis and health state assessment,can we successfully prevent the occurrence of unexpected faults,improve reliability,lower operational costs,and ensure efficient and safe operation of equipment.The Ensemble Empirical Mode Decomposition(EEMD)can decompose the signal into the sum of several Intrinsic Mode Functions(IMF)and the remaining items to reflect the inherent information properties of the signal.Mahalanobis Taguchi System(MTS)is a kind of pattern recognition method which integrates Mahalanobis distance,orthogonal array,and signal-to-noise ratio to classify and diagnose.It has a unique classification and dimensionality reduction theory.This dissertation combines EEMD and MTS,introducing them into the field of health management and exploring in detail the application of EEMD-MTS model in fault diagnosis and health assessment for mechanical equipment.The dissertation includes the following aspects:1)Research on health management systemUpon introducing relevaht theories on Prognostics and Health Management and highlighting their limitations,this dissertation elaborates on the definition and characteristics of mechanical equipment and identifies rolling bearing as the object of study.The health management system for key mechanical parts of is subsequently analyzed and described based on the four key technologies of feature extraction,fault diagnosis,health status assessment and information fusion.The descriptions not only introduce the definitions and functions of various technical methods,but also show the connection and difference between them.2)Research on fault diagnosis based on EEMD-MTSIn view of the shortcomings of the traditional MTS in using orthogonal array and signal-to-noise ratio to filter feature variables,a new method based on rough set is proposed and utilized to resolve the problems in fault diagnosis.Firstly,the vibration signal is decomposed by EEMD and the parameters are extracted as the eigenvector to be input into the MTS.The reference space is then constructed and validated.The rough set is used to replace the orthogonal table and the signal-to-noise ratio to optimize and filter the indexes,realizing the fault diagnosis of the mechanical equipment.This example of fault diagnosis proves that this method is able to effectively determine the fault type of rolling bearing.3)Research on information fusion based on EEMD-MTSHere,it is not only necessary to consider the fault diagnosis for key components of mechanical equipment,but also the fact that mechanical equipment contains various parts and components.Therefore,this dissertation adopts the evidence theory and combines all information collected by various sensor to comprehensively reflect the fault condition of mechanical equipment.Based on fault diagnosis using EEMD-MTS,this method constructs the reference space for each fault mode to establish the basic trust allocation function.The fault diagnosis of information fusion is completed according to the synthesis rules of evidence theory.Finally,the fault case study shows that the method of information fusion can greatly improve the accuracy and reliability of fault diagnosis for mechanical equipment.4)Research on health state assessment based on EEMD-MTSIn order to determine the change pattern for mechanical equipment in time,a health state assessment method based on EEMD-MTS is proposed in view of the lack of relevant research.This dissertation summarizes the criteria for the assessment of health status and only uses the data of normal state to train the model of MTS.The health state assessment model with a full life cycle is subsequently established.A health index is proposed to facilitate the observation of the change process for the whole life cycle.Finally,the effectiveness of the method is proven with a case study of accelerated life test on rolling bearing.