X. Ren, M. Ong, G. Allan, V. Kadirkamanathan, H. A. Thompson and P. J. Fleming

1
NCAF Sunderland Meeting09/06/2004
Integrated Fault Diagnostics on the Grid

• X. Ren, M. Ong, G. Allan, V. Kadirkamanathan, H.
  A. Thompson and P. J. Fleming
• Rolls-Royce UTC in Control and Systems
  Engineering
• Department of Automatic Control and Systems
  Engineering
• University of Sheffield
• UK

2
Outline

• Model-based fault diagnosis
• Service-oriented architecture for Integration
• DAME project
• An engine simulation Grid service
• Case-based reasoning for fault analysis
• Security Issues
• Conclusions and future work

3
Model-based Fault Diagnosis

• Basic idea the measurements from dissimilar
  sensors are functionally related. Any violation
  of these relationships indicates the occurrence
  of faults.

4
Model-based Fault Detection and Isolation (FDI)
Evaluation Criteria
Different Approaches

• Promptness of fault detection
• Sensitivity to incipient faults
• False alarm rate and missed fault detection
• Incorrect fault identification

• Observer approach
• Parameter estimation
• Statistical approach
• Qualitative method

5
How to Improve?

• Adaptive fault-detective threshold
• Likelihood Ratio Test
• Robust approaches
• Soft computing approaches
• Combination of quantitative and qualitative
  methods

6
Necessity of Integration

• Different FDI algorithms have been developed
• Each approach has its own focus no silver
  bullet
• Integration of several FDI schemes will provide
  better decision support for fault diagnosis and
  condition-based maintenance
• HPC power is often a burden for implementations
  of different algorithms
• Large data storage capability are needed for
  online/offline monitoring and fault analysis
• Web Services and service-oriented architecture
  (SOA) for integration

7
Integration of FDI Tools based on SOA

• Service-oriented architecture (SOA)
• An architecture made up of components and
  interconnections
• Designed to stress interoperability and location
  transparency
• Examples include DCOM, CORBA and Web services
• Web services are the most promising techniques
  for SOA Implementation on the Internet
• Structure of Web services

• Different FDI approaches can be wrapped and
  provided as Web services on the Internet

8
Integration of FDI Tools on the Grid

• Grid a distributed computing infrastructure that
  provides dependable, consistent, pervasive and
  inexpensive access to high-end computational
  capabilities
• Coordinated resource sharing and problem solving
  in dynamic, multi-institutional virtual
  organisations -- Ian Foster
• The Grid provides the potential HPC power and
  large data storage capability for implementation
  of different algorithms
• Service-Oriented Architecture and the Open Grid
  Services Architecture (OGSA) for the integration

9
DAME Project

• Distributed Aircraft Maintenance Environment
• An UK e-Science Pilot Project demonstrating the
  use of the Grid to implement a distributed
  decision support system for maintenance

Academic and Industrial Partners
Web Link http//www.cs.york.ac.uk/dame
10
DAME Architectural Overview
DAMEGrid Portal
...
Case Based
Modelling/
Signal
QUOTE
Reasoning
Grid Services Management
Simulation
Processing
Novel
Data
The Grid
l
a
e
n
Data-Mining
c
s
a
a
o
i
t
a
t
i
e
t
v
r
t
t
a
n
a
a
a
a
w
r
a
i
t
r
e
D
D
D
a
P
e
g
a
S
p
R
n
D
O
E
Vibration
Shaft Speed
Fuel Flow
11
Sheffield Project Goals

• Simulation, visualisation and decision support
  environment for engine fault diagnosis and
  maintenance
• Integration of modelling, estimation and analysis
  tools for engine fault diagnostics

12
Engine Simulation as a Grid Service

• An Grid service for engine simulation based on
  the Rolls-Royce engine performance model
• Based on the Globus Toolkit 3 (GT3) and the Open
  Grid Service Infrastructure (OGSI) for
  distributed systems integration
• The engine simulation Grid service can be invoked
  simultaneously in different Virtual
  Organisations for different applications

13
Features of the Engine Simulation Grid Service

• The simulation
• can simulate for different flight operational
  conditions and requirements, e.g. Idle, Take-off,
  Climb, Cruise
• takes different variables, including Net Thrust,
  Turbine Temperature, Spool Speed and Fuel Flow,
  as control input
• As Grid services
• GSI enabled secure engine performance simulation
  via Internet
• a bunch of simulation instances generated by the
  factory service different engine simulation can
  be performed simultaneously
• Only lightweight computers or PDAs are needed to
  perform the engine simulation, while still
  benefit from high-performance/high-throughput
  computing power provided by the Grid

14
Access the Grid Service through Web Browsers

• A Web server for accessing the Engine Simulation
  Grid Service has been developed
• Use JSP to invoke the simulation service
• Use Web browser to view the engine performance
  data and the simulation result
• Two way SSL authentication via server and client
  certificates

15
Applications

• A simulation-based fault detection can be
  performed
• Experienced maintenance engineers can use this
  service to find the problem via on-line
  simulation
• Can be used to confirm the fault assumption
• Can be invoked by different fault diagnostic
  algorithms

Feature Detection and Event Generation
16
CBR for Fault Analysis

• Emulates the diagnostic skill of an experienced
  maintenance engineer
• Provides the user with best practice advice
  when confronted with a set of fault symptoms
• Provides a confidence measure for each suggested
  solution
• Facilitates the processing of logistic data,
  current and historical fault data to isolate
  engine problems

17
Case-Based Reasoning

• CBR is a mature, low-risk subfield of AI
• Primary knowledge source
• a memory of stored cases recording specific prior
  episodes
• not generalised rules
• New solutions generated by adapting relevant
  cases from memory to suit new situations

Retrieve
Propose Solution
Adapt
Justify
Criticize
Reasoning by remembering, reasoning is
remembered.
Evaluate
Store
18
Cases

• Cases represent
• Engine information
• Individual engine fault events
• Explicit fault symptoms and patterns
• Maintenance actions

• Cases link together the knowledge about
• Information concerning the fault
• Response to the fault
• Effects of those responses

19
CBR Maintenance Advisor

• CBR Maintenance Advisor as a Grid Service
• Based on Globus 3 Toolkit
• Access to knowledge gained from previous fault
  diagnosis
• Obtain maintenance advice in form of appropriate
  maintenance action

20
Decision Support for DAME
21
Security Issues

• The secure sensitive engine model and user data
  need to be protected!

• Challenges
• Eavesdropping
• Tampering
• Impersonation

• Current solutions
• Symmetric-Key encryption
• Public-Key encryption
• Digital Signatures using PKI
• Certificate-Based Authentication

• The Engine Simulation Grid Service uses GSI to
  secure the model and user data
• SSL is used to secure the Web and database server

22
GSI and SSL Implementation

• Grid Security Infrastructure(GSI)
• the security functionality implemented by Globus
  Toolkit
• Secure Socket Layer(SSL)
• a technology for secure connection between Web
  browsers and servers.
• Both use certificates for authentication
• SSL uses public key encryption to generate shared
  secrets then performs the symmetric encryption
• GSI provides the Transport Layer Security and
  Message Level Security
• Transport Layer Security a Grid version of SSL
• Message Level Security based on WS-Security, XML
  Encryption and XML Signature standards

23
Conclusions and Future Work

• Importance of integration for fault diagnostics
• SOA implemented on the Grid for FDI integration
• The Engine Simulation Grid Service
• CBR for fault analysis and maintenance
• Performance-based engine fault diagnosis
• More Grid services for engine fault diagnosis
• Knowledge-based workflow advising