Information Technology Reference
In-Depth Information
It ensures that the time period or distance considered are equal in both the target
case and the cases in the case base in order to facilitate similarity calculation.
2.
Single point attributes
The single point attributes represent the flow data measured by a single sensor. These
attributes are calculated for each sensor within time interval
T
and over a length of
distance
L
. In the target case, the starting sensor is at a location and time for which
the accident risk is to be analysed, the current sensor. For the cases in the case base,
usually the starting sensor immediately preceding the location (or time) of an accident
is chosen. The single point attributes, derived from the sensor measurements, include:
•
Average number of vehicles (flow) over all lanes,
f
•
Average velocity of vehicles over all lanes,
s
•
Average headway between vehicles over all lanes,
h
•
Average occupancy measured by the sensor,
o
•
Standard deviation in average number of vehicles per lane (flow) between lanes,
f
d
•
Standard deviation in average velocity of vehicles between lanes,
s
d
•
Standard deviation in average headway of vehicles between all lanes,
h
d
Standard deviation in occupancy of vehicles between all lanes,
o
d
•
3.
Temporal attributes
These attributes describe the variation in the attributes of traffic flow data over the
specified time period
T
[5]. The temporal variations are determined by calculating the
ratio of the difference between the 90th and 50th percentile of the traffic flow
attribute to the maximum value of the attribute found in all lanes. They include the
variation in average speed and flow over all lanes. The variation of the other two
attributes (headway and occupancy) was not considered to be relevant.
•
Variation in average speed over all lanes
V
T,s
,
(1)
where
s
90
, s
50
and
s
max
are the 90
th
percentile, 50
th
percentile and maximum average
speed found over all lanes, respectively, during the time period
T
.
•
Variation in average flow over all lanes
,
(2)
where
f
90
,
f
50
and
f
max
are the 90
th
percentile, 50
th
percentile and maximum average
number of vehicles found over all lanes, respectively, during
T
.
