Automatic Machinery Fault Detection and Diagnosis Using Fuzzy Logic
Abstract
Machine condition monitoring incorporates a number of machinery fault diagnosis techniques. Many of these machinery fault diagnostic techniques involve automatic signal classification. In this paper Fuzzy logic techniques have been applied to classify frequency spectra presenting various bearing faults. The frequency spectra have been processed by four common Fuzzy set shapes: linear, triangular, S-curve and Pi curve. The application of basic Fuzzy logic techniques has allowed Fuzzy numbers to be generated which represent the similarity between two frequency spectra. Correct classification of six different bearing fault spectra was observed when the frequency spectra were represented by Pi curves. The degree of membership of each individual spectrum with respect to the other spectra, however, indicated a certain degree of overlapping. Further investigations must be conducted in order to optimize the ability to classify spectra with a certain degree of overlapping or masking.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1996
- Accession Number
- ADP010161
Entities
People
- C. K. Mechefske
- D. Prendergast
- J. Del Mar