Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.523970
Title: Any time probabilistic sensor validation
Author: Ibarguengoytia, P. H.
Awarding Body: University of Salford
Current Institution: University of Salford
Date of Award: 1997
Availability of Full Text:
Access through EThOS:
Access through Institution:
Abstract:
Many applications of computing, such as those in medicine and the control of manufacturing and power plants, utilize sensors to obtain information. Unfortunately, sensors are prone to failures. Even with the most sophisticated instruments and control systems, a decision based on faulty data could lead to disaster. This thesis develops a new approach to sensor validation. The thesis proposes a layered approach to the use of sensor information where the lowest layer validates sensors and provides information to the higher layers that model the process. The approach begins with a Bayesian network that defines the dependencies between the sensors in the process. Probabilistic propagation is used to estimate the value of a sensor based on its related sensors. If this estimated value differs from the actual value, then a potential fault is detected. The fault is only potential since it may be that the estimated value was based on a faulty reading. This process can be repeated for all the sensors resulting in a set of potentially faulty sensors. The real faults are isolated from the apparent ones by using a lemma whose proof is based on the properties of a Markov blanket. In order to perform in a real time environment, an any time version of the algorithm has been developed. That is, the quality of the answer returned by the algorithm improves continuously with time. The approach is compared and contrasted with other methods of sensor validation and an empirical evaluation of the sensor validation algorithm is carried out. The empirical evaluation presents the results obtained when the algorithm is applied to the validation of temperature sensors in a gas turbine of a power plant.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.523970  DOI: Not available
Keywords: QA075 Electronic computers. Computer science ; QA Mathematics
Share: