Recurrence Complex Network-based Fault Recognition Of Rotating Machinery

With the rapid development of science and technology level,the development of mechanical equipment is more and more complex,and the contaction between components is becoming more and more closely.The closely cooperation between components brings the advantages of high efficiency industrial production.If one component fails,the results will be worse.Which not only affects the production efficiency,but also brings incalculable economic loss.Generally there are three main steps of fault diagnosis.Firstly,acquiring vibration signals from working mechanical equipment.Secondly,fault feature extraction.Finally,working condition recognition.The second step is the core which was deeply studied in this paper.The construction of the recurrence complex network was deeply studied,and the mutual information method and Cao method ware introduced to determine the phase space construction parameters.Best delay time ? and embedding dimension m maitain many attractor properties of the recurrence complex network,so as to vibration signal de-noising and fault diagnosis.The local projection noise elimination method was applied to rolling bearing vibration signal de-noising.The effectiveness of the proposed method had been verified through simulated signals and experimental data.In this paper,a fault diagnosis method for rotating machinery based on integrating local projection de-noising and recurrence quantification analysis method was proposed.Selecting rolling fault simulation test bench as the research object,the signal acquisition system was build based on LabVIEW,different types of rolling bearing fault signal were abtained.Firstly,the vibration signal de-noising by local projection method.Secondly,the recurrence plot was drew,and then acquire features reflected on recurrence plot by recurrence quantification analysis method.Determinism and entropy ware selected as feature vector.Finally,feature vector sets of training samples were clustered by Kernel Fuzzy C-Means(KFCM),clustering centers of different signal types were obtained,the principle of the minimum Euclidean distance was adopted as the recognition method of feature vector sets of testing samples In the same situation,the statistical characteristics of recurrence complex networks ware used in fault diagnosis of rolling bearing,the superiority of recurrence quantification analysis was verified.Finally,selecting hydraulic pump fault simulation test bench as the research object,diagnosis of different types fault signal were achieved.the results showed that the effectiveness of the proposed method in rotating machinery fault recognition.