Development of an Information Monitoring and Diagnostic System
Diagnostic Technology and System Design Criteria
Another important task in Phase 1 was to investigate and evaluate
diagnostic methods, tools, and techniques for inclusion in the
current project. We conducted a broad review of possible approaches
for diagnostics and determine the degree of technical maturity
with which each has been applied to building problems. We defined
a set of criteria and then evaluated options in terms of these
criteria. Our analysis considered issues such as required sensor
and communications technology, bottom-up versus top-down diagnostics
architecture, and the design of temporary versus permanent systems.
We also examined the status of techniques from the field of intelligent
systems (e.g., artificial intelligence, fuzzy logic, neural networks)
and diagnostics used in process control industries. A diagnostic
system comprises the components depicted in Figure 2.
Figure 2. A depiction of the components of a diagnostic system
There are many approaches for each of the components, so one must
define a set of criteria and evaluate the suitability of the approaches.
We used the following six criteria:
- Diagnoses must be understandable to building technical
personnel. In other words, the resulting system must make diagnoses
using a process that building engineers understand (perhaps after
some training).
- Resulting diagnoses generated by the system must be believable
to building engineers. In order to act on the diagnoses, building
engineers must make recommendations to upper management under
tight financial constraints. Given the career risks involved,
this process will only be undertaken if the diagnostic system
presents credible evidence of its findings.
- Diagnostic approaches must be robust in the sense that
they can be gainfully applied to a specific building without an
enormous amount of tuning. This can either be the result of extensive
work by others certifying that a technique works on a large group
of building applications (including the specific test building)
or it could be the result of some inherent property of the technique
that grants it very broad and easily defended applicability to
a large group of buildings (including the specific test building.
- Diagnostic components for individual building components must
coexist, cooperate and integrate in order to a) be able
to work together on separate systems within a building and b)
to permit expansion of the system to other building components.
- Diagnostic approaches must be easy to implement at reasonable
cost.
- The design of the initial diagnostic system must lead naturally
to increased automation in later phases.
The overall architecture of diagnostic systems includes:
- bottom-up and top-down approaches
- human fault detector, automatic fault detector, or hybrid
methods
- temporary, short-term systems or permanently installed systems
We concluded that for that for the target Class A office buildings,
a top-down architecture is promising because it is integrates
better and costs less to design than bottom-up systems. Bottom
up systems detection of performance failures associated with specific
individual devices assuming a fixed range of operating conditions
(in the face of the great diversity of conditions found in real-world
applications) (Hyvarinen and Kohonen, 1993, Hyvarinen, and Karki,
1996). Similarly, human assistance in fault detection appears
more promising in the near term. Increased automation is a viable
longer term strategy as data collection is improved and automated
techniques will be easier to build. Automated techniques require
training statistical models with data sets that are limited by
current building monitoring systems. After building the models
one must test their ability to detect various categories of faults.
Continue to IMDS Design
Return to the Table of Contents.
Return to Technology Innovation and Adoption Theory.
