Track 2 Session 5
9:10 to 10:10 a.m. Thursday March 26, 2009

Operational Reliability: How Well Do You Know the Behaviour of Your Product/System?

This presentation shows how data mining, neural networks and fuzzy logic can be integrated into maintenance activities in order to generate and manage knowledge with respect to equipment/system behaviour in relation to its operating conditions. It demonstrates how this knowledge has proven invaluable to achieve improvements in system operational reliability and availability by: 1) providing design engineers with additional knowledge for future equipment design and at the testing and validation stages of development; and 2) providing maintenance engineers with the knowledge and ability to predict "normal" system behaviour in different operating environments, and therefore design diagnostic tools to detect any deviation from such behaviour. A case example is presented in which it is demonstrated how data mining and intelligent diagnostic techniques have brought benefits to the maintenance of a high-speed (250km/h) train fleet based in Madrid and subsequent improvements to in-service reliability.