Development of Automated Condition Monitoring - Diagnostics / Prognostics using AI Tools

The program will focus on multiple sensor data sets, which include known faults, in order to develop automated diagnostic / prognostic capabilities. Currently data sets are available from University of Southampton in-house bearing rig testing and include a range of sensor data such as oil temperatures and pressures, shaft speed, bearing load, various oil debris monitoring sensors, vibration etc. The program should investigate the application of AI tools for fault diagnostic / prognostic. Factors should include identification of primary and secondary sensors for known fault types, the effect of missing / poor data, anomaly detection, temporal association, long term trending etc. There will be the opportunity to undertake further rig testing in order to develop, verify and validate the approaches taken.

Smiths Aerospace has developed in-house AI tools for advanced data manipulation based on Data Mining (cluster model / decision tree / neural nets / association rules) and Causal Network Reasoning for data fusion and ultimately automated diagnostics. This will form part of the key toolset used in the proposed PhD program to develop automated condition monitoring.

The programme, funded by Smiths Aerospace, will be supervised by Professor Robert Wood from the Materials and Surface Engineering Research Group within the School of Engineering Sciences.

Students interested in this unique opportunity and expecting to graduate or have graduated with a degree in Maths or Physical Sciences or Engineering Sciences and have a suitable analytical background, should complete the PhD studentship application form found at:
www.soton.ac.uk/Prospectuses/Admissions/ApplyingtoSouthampton/ApplicationForms//
and can contact Professor Robert JK Wood (T: 02380 594881, E: rjw3@soton.ac.uk) for further details. A very competitive maintenance grant will be available.

To be returned no later than 29 Jul 2005. Please quote reference number 04R0826