Grid-based on-line aeroengine diagnostics The DAME pilot project Jim Austin, University of York The Problem • Information is being gathered at growing rates at an international level • This information needs to be fully utilized • Technology is needed to integrate and exploit the data • Grid is a part of the answer to this 2 Aims • To build a distributed, Grid based, diagnostic maintenance system • To prove the technology on a Rolls Royce Aeroengine diagnostic maintenance problem • Demonstrate the process of building a Grid based system • To deliver grid-enabled technologies that underpin the application 3 The Application • The support of engine diagnostics on a global scale. 4 Engine flight data London Airport Airline office New York Airport Grid Diagnostics centre Maintenance Centre American data center European data center 5 Aircraft Engine Outline architecture Quote Systems Diagnostic operator Engine data Engine data log AURA data search Operational Report Model-based Interpretation Maintenance operator Database of Operational data Case-based Reasoning Decisionsupport 6 Example use case diagram GRID Environment Local Environment <<extend>> Data Engine Perform Extended Analysis / Modelling Diagnosis / Prognosis Status / Parameters Perform Analysis Update Operation Provide Domain Expert Assessment Update Local Diagnostics Diagnosis Assessment Inform Domain Expert of Undetected Problem Results Reports Information Results Maintenance Team Domain Expert / Maintenance Planner Store Result of Diagnosis and Operation Provide Statistics Report 7 Challenges • Support on-line diagnostics in real time • Deal with the data from 100,000 engines in operation • Prove pattern matching methodology • Prove the business case for the technology 8 Technologies • AURA: High performance search technology • QUOTE: On-engine diagnostics system • Globus: Grid software • WR Grid: Demonstrator hardware 9 AURA • High performance ‘search engine’ • Based on neural networks • Develop for distributed operation 10 QUOTE • • • • On-engine health assessment Under trials on Trent 500 now Will identify novelty Some diagnostics 11 12 White Rose Computational Infrastructure Oxford Super Janet Leeds Cluster Leeds Shared Memory White Rose Computational Grid Sheffield Distributed Memory York Shared Memory 13 Our developing architecture 14 Industry Collaborators • DS&S : Rolls Royce data services providers • Rolls Royce : Data and problem • WRCG: Esteem, Sun and Streamline: Demonstrator Grid • Cybula: AURA technology support 15 Academic team • Austin – project management, data management • Tarassenko, Austin – algorithms for fault identification • Dew, Djemame – system architecture • Fleming, Thompson – decision support • McKay – data modeling 16 Academic Team • McDermid – Dependability • Wellings – Real time issues Researchers - 15, including... • Tom Jackson - Coordinator • Martyn Fletcher - Software Manager 17 End 18