|
P2-5. Pattern
Recognition Based Fault Detection and Diagnostics > Approach
Objectives:
Develop pattern-recognition techniques that
automate the detection and diagnosis of faults
based on comparison of plots of good and
bad system performance, obtained from both
simulation and actual field measurements
and stored in a library for a broad number
of faults. A commercially available diagnostic
tool often used in building commissioning
that makes use of this approach, but with
expert interpretation applied manually, is
Architectural Energy Corporation's ENFORMA.
This project will extend this technique by
taking the additional step of automating
the interpretation step and presenting conclusions
to the user.
Approach:
The steps taken to
perform this project
were
the following:
- Examine, categorize, select, and document
initial set of diagnostics to be automated.
- Review, select and document pattern recognition
techniques to be applied.
- Implement and test initial pattern recognition
algorithms in software.
- Develop a user interface within the framework
of the Whole Building
Diagnostician Developer's
Toolkit or a stand-alone
interface to demonstrate
the initial pattern
recognition algorithms.
- Perform field test and document results.
- Develop a software module compatible with
the Whole Building Diagnostician, or generic
code suitable for adaptation in other stand-alone
software programs. Expand the user interface
to accommodate improvements based on the
field test under Task 2.5.5.Project 2.5 Outcomes.
Back to Previous Page
|
