This thesis is concerned with extending process monitoring and diagnosis to electrical and mechanical utilities. The motivation is that the reliability, safety and energy efficiency of industrial processes increasingly depend on the condition of the electrical supply and the electrical and mechanical equipment in the process. To enable the integration of electrical and mechanical measurements in the analysis of process disturbances, this thesis develops four new signal analysis methods for transient disturbances, and for measurements with different sampling rates. Transient disturbances are considered because the electrical utility is mostly affected by events of a transient nature. Different sampling rates are considered because process measurements are commonly sampled at intervals in the order of seconds, while electrical and mechanical measurements are commonly sampled with millisecond intervals. Three of the methods detect transient disturbances. Each method progressively improves or extends the applicability of the previous method. Specifically, the first detection method does univariate analysis, the second method extends the analysis to a multivariate data set, and the third method extends the multivariate analysis to measurements with different sampling rates. The fourth method developed removes the transient disturbances from the time series of oscillatory measurements. The motivation is that the analysis of oscillatory disturbances can be affected by transient disturbances. The methods were developed and tested on experimental and industrial data sets obtained during industrial placements with ABB Corporate Research Center, Kraków, Poland and ABB Oil, Gas and Petrochemicals, Oslo, Norway. The concluding chapters of the thesis discuss the merits and limitations of each method, and present three directions for future research. The ideas should contribute further to the extension of process monitoring and diagnosis to the electrical and mechanical utilities. The ideas are exemplified on the case studies and shown to be promising directions for future research.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:656565 |
| Date | January 2014 |
| Creators | Cecílio, Inês M. |
| Contributors | Thornhill, Nina F. |
| Publisher | Imperial College London |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/10044/1/24962 |
Page generated in 0.0015 seconds