Font Size: a A A

Theory Of Granular Computing And Its Application In Mechanical Fault Diagnosis

Posted on:2009-10-23Degree:MasterType:Thesis
Country:ChinaCandidate:X R ZhuFull Text:PDF
GTID:2178360245456681Subject:Mechanical Manufacturing and Automation
Abstract/Summary:PDF Full Text Request
Granular computing theory is a new concept and calculation model of information processing. It is a hot issue in the field of artificial intelligence, and becomes an important tool of processing massive, incomplete, imprecise, and illegible information. Granular computing theory covers most of the concept and research concerns granular. As a filiation of soft-computing, it combines computing with words theory, quotient space theory, and rough sets theory. This research mainly studied the use of Granular computing theory and quotient space theory in diagnosing rotary machinery fault. There are numerous data of diagnosing rotary machinery fault accumulated during the development of this technique. By analyzing those data, this research aimed to differentiate and regulate the fault diagnosis using the granular computing theory. Main contributions are:1) Introduce the basis concept of granular computing theory; describe the three main granular computing models and their relationships; emphasize the rough sets theory and quotient space theory; point out some fundamental questions of granular computing.2) Design an revised division discretization calculation. This revised method decide the partition distance based on the data itself, in order to reduce the error caused by discretizing rotary machinery decision table.3) Study the fault diagnosing regulation based on rough sets theory, mainly research the attribute reducation of machine malfunction. Point out the limitation of attribute reducation computing from a granular perspective, and improve this computing by using the feature correlation index concept.4) Based on the result of quotient space theory identification of rotary faults, the time-domain wave of various fault are significantly different at rough granularity space, so we could distinguish the different fault by specific region wave, and this research used quotient [7] as the fault quantitative feature, provide a new method of quantitative feature.Granular computing theory is a developing theory, future research should focus on better usage in fault diagnosis.
Keywords/Search Tags:Fault Diagnosis, Granular Computing, Discretizing Attribution, Attribute Reducation, Feature Extraction
PDF Full Text Request
Related items