UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Fault isolation in complex dynamic systems using artificial intelligence techniques

Barac, Jovan; (1990) Fault isolation in complex dynamic systems using artificial intelligence techniques. Doctoral thesis (Ph.D), UCL (University College London). Green open access

[img] Text
Fault_isolation_in_complex_dyn.pdf

Download (6MB)

Abstract

Complex dynamic systems, such as computer networks are characterised by the occurrence of multiple faults, sensor faults and intermittent faults as well as single faults. The aim of this dissertation is to provide a general, system independent expert system shell for isolating faults in complex dynamic systems. First a new high level architecture suitable for fault isolation in complex dynamic systems is suggested. This architecture is based on the coupling of a blackboard system with an Assumption-based Truth Maintenance System (ATMS). The blackboard provides a flexible means of control for the inferencing process, while the ATMS provides the means to store and retrieve the data, a record of the inferences, support for the blackboard functions and general support for non-monotonic and multiple context reasoning. Then a new fault model is developed which combines a unified approach to single, multiple and sensor faults with a simple temporal fault logic for intermittent faults. Following that, a novel, algorithmic test optimisation mechanism, based on the calculation of the expected entropy of each test is developed. This mechanism is used to determine the next test that should be performed and completes the system independent elements of the proposed diagnostic environment. The subsequent section shows how the fault model, the test optimisation algorithm and the high level architecture are combined to provide a diagnostic shell and how the ATMS and the blackboard cooperate in solving a diagnostic problem. The final section discusses the implementation, the performance evaluation, the preliminary validation in the chosen domain of computer networks and future validation and research issues.

Type: Thesis (Doctoral)
Qualification: Ph.D
Title: Fault isolation in complex dynamic systems using artificial intelligence techniques
Open access status: An open access version is available from UCL Discovery
Language: English
Additional information: Thesis digitised by ProQuest.
Keywords: Applied sciences; Fault detection
URI: https://discovery.ucl.ac.uk/id/eprint/10107663
Downloads since deposit
15Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item