Use this URL to cite or link to this record in EThOS:
Title: The application of advanced signal processing techniques to condition monitoring
Author: Murray, Angus
Awarding Body: University of Aberdeen
Current Institution: University of Aberdeen
Date of Award: 1997
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
The application of Artificial Neural Networks to condition monitoring has been shown to offer substantial rewards, both in improving safety and reducing maintenance costs. However, introducing ANN's to real industrial situations, highlights a need to improve the application of this powerful technology. This thesis attempts to highlight some of the issues associated with applying ANN's, by first considering how to diagnose combinations of faults. It is shown that by applying a modular approach, using one network per fault, a means is available to accommodate combinations of faults, effectively diagnosing previously unseen data. However, fundamental to the application of ANN's for Condition Monitoring is the improvement of the signal pre-processing which needs to be performed. Traditionally Fourier analysis plays a major role in vibration analysis, constructing a signal from harmonic components. However, there are problems associated with this technique, and with the implementation of ANN's comes a requirement to investigate alternatives. The Wavelet Transform is investigated, not necessarily as a replacement for Fourier analysis, but to complement the technique. With this relatively new transform, comes a means of multi-resolution analysis, capable of providing time-localised information. It is shown that with this multi-resolution, time-frequency approach, a greater understanding can be gained of vibration data. Fourier analysis however still plays a major role in condition monitoring, and so is considered, building on this technique by introducing Higher Order Spectra (HOS). With HOS comes the ability to highlight frequency coupling which may exist within signals, while also greatly reducing noise. The use of HOS for condition monitoring is demonstrated, with interesting higher order properties being associated with faults. And by considering combinations of fault conditions, it is shown how higher order components are generated which are not present in the isolated faults.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available