Fault Diagnosis Of Rolling Bearing Based On EMD |
Posted on:2013-12-22 | Degree:Master | Type:Thesis |
Country:China | Candidate:H Shao | Full Text:PDF |
GTID:2232330395957018 | Subject:Mechanical Manufacturing and Automation |
Abstract/Summary: | PDF Full Text Request |
The rolling bearing is the pivotal and damageable part of the rotary machines, Sofault monitoring and diagnosis of the rolling bearing is a very concerned subject in thefields of engineering and technology both at home and abroad. A large number ofresearches prove that vibration signal analysis is the most efficient method so far.Normally, the vibration signal is nonstationary and nonlinear. EMD(Empirical ModeDecomposition)is self-adapting, especially suitable for the decomposition of thenonstationary and nonlinear signals.This paper introduces the fault modes of the rolling bearing and theircorresponding failure mechanism. The characteristic frequencies and naturalfrequencies are calculated. Natural frequencies of each fault mode are got by the timefrequency analysis of the vibration signals. Tentative diagnosis is made based on thecomparison between the theoretical calculations and the simulated results.This paper applies the superiority of accurate band allocation of wavelet to havethe vibration signals de-noised. The TI(Translation-invariant)method is used in signalde-noising and performs good. Then IMFs(intr ins ic mode functions) are got after thesignals are decomposed using the EMD method. The energy of the first7IMFs arecalculated to form feature vectors because the characteristics are contained mostly in thefirst several IMFs. BP neural network and SVM(support vector machine) are both usedfor the diagnosis of fault modes of the rolling bearing and the comparison is madebetween the two effects of recognition of the two methods. |
Keywords/Search Tags: | Rolling Bearing, Fault Diagnosis, Wavelet, Empirical ModeDecomposition, Neural Network, Support Vector Machine |
PDF Full Text Request |
Related items |