Font Size: a A A

Fault Prognosis Method And Application Of Machinery Buseii On ISOMAP

Posted on:2013-08-16Degree:MasterType:Thesis
Country:ChinaCandidate:H YinFull Text:PDF
GTID:2232330374975914Subject:Vehicle Engineering
Abstract/Summary:PDF Full Text Request
The research of intelligent mechanical fault diagnosis method has been a hot issue for thefield of machinery diagnosis. With the rapid development of the artificial intelligence,computer software technology, modern sensor technology and modern signal processingtechnology, the dimension of the fault signal also will be getting higher and higher and thefault signal data set of large-scale machinery and equipment is a large amount of data, highdimensionality, random regularity. As much as possible to guarantee the geometricrelationships between the datas, we can map the high-dimensional space into low-dimensionalspace manifold, which not only can reduce computational, but also can identify key features,remove noise interference, and comprehensively improve the efficiency of fault diagnosis.Because the characteristics of mechanical failure signal is high dimensionality, thetime-varying, nonlinear and in non-Gaussian distribution, this paper proposes the improvedISOMAP algorithm to the nonlinear dimensionality reduction of mechanical failures signal,making failure data more easily classified. After analysis the principle and calculation processof the classical ISOMAP algorithm, find the short comings of the algorithm when it is appliedin mechanical fault diagnosis. Then we proposed a supervised and fast ISOMAP algorithm formechanical fault diagnosis. Use this improved algorithm for the nonlinear dimensionalityreduction of failure data, we can get a low-dimensional structure. Then we apply supportvector machines to realize fault diagnosis and classification. The main work is(1)Explore the problems of the classical ISOMAP algorithm used in the field ofmachinery fault diagnosis(2)Facing the existing problems of ISOMAP algorithm application to mechanicaldiagnose, propose the supervised ISOMAP algorithm with fast computing speed.(3)Do some gearbox experiment on the automotive transmission test bed to prove theeffectiveness and superiority of this supervised and fast ISOMAP in the gear fault diagnosis.The improved ISOMAP algorithm can effectively reduce dimensions of the fault dataand identify intrinsic dimension, which will greatly reduce the computation time, and improvethe efficiency of fault diagnosis and the correct rate.
Keywords/Search Tags:ISOMAP, Data Dimensionality Reduction, Supervised Learning, MechanicalFault Diagnosis
PDF Full Text Request
Related items