Research On Fault Diagnosis Method Of Rolling Bearing Based On IFD And

In recent years, with the rapid development of Chinese technology and acceleration of new industrialization course, the application of large-scale mechanization becomes more popular. And the degree of automation and information is getting higher and tends to be intelligent, which strongly pushes the economic and social development. However, machinery’s unplanned downtime and local fault, in the industrial production and application, may directly lead to the loss of productivity effect and huge economic losses, even to safety accidents. Roll bearing, as the key part of rotating machinery, will directly affect the overall performance of the system. Thus, it is significant to monitor and diagnose its running state.The roll bearing’s faults diagnosisboils down to the process of signal feature extraction and fault recognition. Regarding the existing problems of hard-to-extract faint faults during the early period and low fault type identification rate, this paper proposes the intelligent faultdiagnosticmethod of rolling bearing based on Intrinsic Frequency-scale Decomposition (IFD), Support Vector Machine (SVM) and Least Squares-Support Vector Machine (LS-SVM).Accordingto the characteristics of roll bearing’svibrationsignal, this method could adaptivelyextract the faint fault features to identify the fault type.The main contents of this paper are as follows:First, this paper focuses on a new time-frequency analysismethod, namely, Local MeanDecomposition (LMD), to analyze algorithm principle and make asimulation analysis of computational process. This research emphasizes anexpansion method of instantaneous phase and a solving method of instantaneous frequency. The simulation analysis of LMD will be applied in roll bearing’s fault feature extraction.Based on confusing issues for LMD’s mode, this paper puts forward IFD and carries out a simulation and comparative study between LMD and IFD.Second, regarding the existing problems of hard-to-extract faint faults during the early period and low fault type identification rate, this paper adopts SVM to build a model of fault recognition. In this paper, the authorproposesa faultdiagnosticmethod for roll bearingbased on IFD and neural network. This research uses IFD as a preprocessor for extracting roll bearing’s kurtosis and energy featureparameter, and for constructing the characteristic vector, so as to enter into SVM for fault type discrimination. Through the example analysis of roll bearing’s signals, this method has been proved to be reliable and practical.Third, combining with Least Squares Support Vector Machines (LS-SVM), this paper presents a faultdiagnosticmethod for roll bearingbased on IFD and LS-SVM, and extracts features by applying IFD to constructfault feature vectors.The author applies the above method to automaticallyclassify and identify the roll bearing’s operating condition and fault type. Through the example analysis of roll bearing, this method has been proved to be valid, which provides a new approach to achieve the online intelligentdiagnosis for roll bearingfault.