Instrumentation & Measurement Magazine 25-1 - 59

can be done safely and without affecting the actual building
operations.
The sensor data are collected from simulation and experimentation
on the physical model of the building, while WOs
are collected in the CMMS database (from occupants' calls or
operator's inspections) and are shown in Fig. 4.
Since there are numerous processed datasets available, the
BIM has a vital role in the storage and visualization of information
for use by decision-makers. In practice, when operators
want to extract temperature WO, they must run the CMMS
application and toggle between several programs to process
the raw data, spending a great deal of time to get through
this process which may lead to inaccurate decisions. We automate
this process by transforming the datasets into a set of
maps using Dynamo [10], as shown in Fig. 4. Maintenance
teams can analyze the spatial distribution of WOs at diverse
scales (room, building, or campus) using different metrics (total
WO, WO intensity or intensity change rate of WOs). Such
maps can expedite and facilitate the measurement of the operational
performance of buildings. Changes in the data are
displayed, based on their parameters, and they are reflected in
plans, schedules, and 3D views. For instance, Fig. 4c (top chart)
shows that 80% of all WOs were issued for 19 buildings out of
42 located on the campus. Similarly, the building-level results
(bottom chart) allow building operators to focus on the spaces
with more corrective orders or complaints in each building.
This way, the performance and fault patterns of the research lab
are identified, based on a distinct data source other than sensor
readings. Also, the quantified maintenance performance
of the studied lab along with the other spaces in the building
are graphically displayed in the model by color-coding objects
and using existing color schemes to represent ranges of numerical
data. The illustrative
outcomes allow the maintenance
teams to analyze
the spatial distribution of
WOs. The raw data can be
processed and transformed
into quantifiable information
[11] and stored in the
cloud to assess the performance
of the sensor fusion
model.
Sensor Fusion
Model (SAFE)
The sensor layer contains
a group of sensors measuring
temperature. SAFE
adopted a competitive sensor
fusion configuration
in which each sensor measures
temperature the same
way and sends its output
to the sensor fusion algorithm,
which examines the
February 2022
values and decides what readings to keep or discard. The algorithm
takes raw measurements from the sensors as inputs,
performs a principal component analysis calculation, and
returns a single output. The output values from the sensor fusion
model are considered correct even if some input sensors
have incorrect readings [8]. This algorithm requires no prior
knowledge, avoids subjectivity in the calculation of correlation
matrices, and its implementation is simple and easy to
understand.
We look for spikes and outliers in the sensor readings to
confirm sensor failures. Also, we check for erratic values which
occur when most of the input sensors to the fusion layer have
different readings.
BIM Integration
The data collected from SAFE is displayed into a BIM using
cloud services [9]. Fig. 5 shows the integration of the data
collected from various sensors, fused temperature, and the
average temperature of the room using SUSTAIN. The application
allows an interface to be built, including a timeline
scrubber (to query data at a certain time), sensor lists (with
temperature values) and a BIM integrated with temperature
readings visually and contextually (using different colors for
temperature values). Fig. 5 shows the sensor values, for instance
the value of Room 3222 T1 is 1.92 °C and Room 3222 T5
is 0 °C, indicating two faults in the system. Each sensor is also
linked to a line chart to show the temperature readings over
time, as seen in Fig. 6, which shows the fused temperature and
the average temperature of the sensor readings.
The average temperature has many variations due to
faulty readings we introduced during experimentation, but
fused temperature provides a steadier temperature value by
Fig. 5. Integration of data with BIM using Autodesk Forge [9].
IEEE Instrumentation & Measurement Magazine
59
Instrumentation & Measurement Magazine 25-1

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-1
