Based On The Characteristics Of LMD And Rotating Machinery Fault Diagnosis Method Of Extraction And Analysis

The original production tools was replaced by machinery since the two industrial, and thenmachinery become an indispensable tool in people’s life. Rotating machinery such as turbines,generators and other large rotating machinery took lots part of our production ifeld gradually. Butwith the speed of rotating machinery increasing, the probability of failure become higher andhigher，so the requirements for the mechanical performance also reach to a high point, especiallyin terms of the stability. In addition, stopping the function of large machinery in order to makemaintenance take a large of cost, so it’s necessary to process fault diagnosis for the equipment.First, the signiifcance of the fault diagnosis for rotating machinery is preferred andintroduces several methods of fault diagnosis, besides points out that the main method used inthis paper is vibration diagnosis. And then enumerates several fault features extraction methodssuch as FFT，wavelet transform, HT and LMD etc.. Finally take the time-frequency analysis asthe main signal processing method by make a comparison between several signal processingmethods.Both of LMD and HHT has their own unique advantages as a time-frequency signal analysismethod. In the second chapter, we mainly study the algorithms of the two methods in detail, andthen make a diagnostic eiffcacy comparison between them, the result show that LMD has moreadvantages than HHT, is the main signal processing method used in this paper.But LMD also has some defects cause it’s a method of experienced. The sampling effect andthe accuracy of the LMD decomposition is researched in fourth chapter, next we make a detailanalysis for the selection of LMD sliding average span and the method of LMD enveloping,according to the simulation show that adopting the appropriate span of the sliding and the rightenvelope method can improve the accuracy of LMD decomposition, restrain the end effecteffectively.Based on the above improved LMD algorithm, author put forward two kinds of new rotatingmachinery extraction methods at last. One is making the statistical analysis for the kurtosis basedon the IF which got by LMD decomposition; Another method is similar to the ifrst method, thedifference is that the kurtosis of method two is based on the distribution function of IF instead ofbased on the IF directly. At last, through the simulation experiments show that the second methoddistinguish the two kinds of fault successfully.