Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.632437
Title: A phased mission approach to fault propagation
Author: Lloyd, Michael D.
Awarding Body: University of Nottingham
Current Institution: University of Nottingham
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
On complex systems with built-in health management systems, the faults diagnosed during a mission can number in the tens of thousands. When these faults are evaluated, many are found to be false. This work has, therefore, developed a technique by which diagnosed faults can be evaluated using known system data and a system modelling technique to automatically verify their legitimacy. Petri nets (PNs) were selected as the modelling technique since they allow systems to be modelled in a componentistic and flexible way, that still provides a high level of accuracy. The PN technique was used to model the performance of an experimental facility, the BAE Systems fuel rig, which represents an aircraft fuel system. A wide range of faults were injected into the system and sensor outputs were recorded. By comparing the sensor outputs from the fuel rig to the PN predicted system behaviour, the faults were assessed as either genuine or false. The standard deviation technique is used as part of the comparison process as it provides a high level of detail with low computational requirements. A piece of software was written to automate the PN simulation and comparison of the output data. The ability of the overall technique to verify diagnosed faults was demonstrated by a thorough consideration of failure modes in the fuel rig system. First and second order faults were evaluated and the results showed that the technique was very successful at identifying both genuine and false faults. Some issues were evident when hidden failures were considered and faults which were revealed for only short periods of time were injected. The PN technique was also successfully used to model the behaviour of the fuel system of the Airbus A340 aircraft. This system contains a higher level of complexity in terms of both design and operation compared to the fuel rig. The behaviour of the system in normal operation was modelled to replicate that described in literature and a number of first and second order faults were modelled. The PN predicted behaviour of the fuel system in the presence of these faults matched well with that expected. The PN technique can be used to obtain the output of sensors when failures occur, and such information can be used in the process of system design. An approach is presented by which a sensors value can be calculated and used to select sensors in a system. The technique considers the change in the value measured by a sensor as a result of faults for single sensors and their pairs.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.632437  DOI: Not available
Keywords: TA Engineering (General). Civil engineering (General)
Share: