Fault diagnosis in unmanned aerial vehicles
- Fast binary decision algorithms enable real-time in-flight diagnosis
Fast and accurate decision making related to in-flight faults in unmanned aerial vehicles is essential if potentially catastrophic mission failure is to be avoided.
Fault Trees provide a diagrammatic description of the various causes of a specified system failure in terms of the failure of its components. Fault Tree Analysis (FTA) is used to predict the failure likelihood of complex safety critical engineering systems. However, FTA is such a computationally intensive method that for large-scale systems, approximations are required.
The Department of Mathematical Sciences has developed an alternative computational method for real-time mission planning, based on Binary Decision Diagrams. This new method overcomes the inherent weakness of the FTA approach by performing the calculations exactly – and is much quicker – offering considerable advantages in efficiency enabling the technique to be used in new contexts.
The underlying work led to an EPSRC-funded project with BAE Systems and 10 other universities as project partners – NECTISE (Network Enabled Capability Through Innovative Systems Engineering). This project aimed to enable defence suppliers such as BAE to develop systems capable of responding to changing requirements in complex dynamically connected networks of supplier customer organisations.
The recognised importance of the Binary Decision Diagrams method led to its inclusion in the highly significant ASTRAEA (Autonomous Systems Technology Related Airborne Evaluation and Assessment) project that focussed on the technologies, systems, facilities, procedures and regulations to allow unmanned aerial vehicles to operate safely and routinely in civil UK airspace.
The research has raised awareness of the advantages of Binary Decision Diagrams, resulting in integration into major industrial trials and proprietary software products, including at BAE Systems.
REAL-TIME DECISION MAKING
Binary Decision Diagrams are used to support decision-making on system utilisation in real-time operation. For example, as component faults are reported or environmental conditions change, fast calculations can be performed to update the prediction of the probability of mission failure so that appropriate action can then be taken.
RESPONDING TO CHANGING CONDITIONS
Unmanned aerial vehicles are being deployed in a wide range of contexts including observation, security, monitoring and law enforcement. The Binary Decision Diagram approach is being used to support the reconfiguration of missions that are at risk of failure due to changing conditions.
NEW PRODUCTS AND SERVICES
Commercial software companies such as ISOGRAPH, ARALIA and ITEM SOFTWARE now incorporate Binary Decision Diagrams in their proprietary packages for Fault Tree Analysis. The research therefore changed the awareness of another industry leading to new products and services.