Information Technology Reference
In-Depth Information
has the highest pedestrian volume with the lowest traffic volume among the four sites.
The length of Site 3 is 2.15 km. It has 11 and 8 access points on direction 1 and direc-
tion 2, respectively. The site has the highest traffic volume (among the four sites) on
both weekday and weekend. Pedestrian volume is also high at this site. Site 4 is 2.94
km long, 4 access points on both directions with no pedestrian crossing. The traffic
volume is moderate with very little pedestrian activity. The traffic and pedestrian
count data were converted to hourly volume data prior to developing the FLM. It is to
be noted that all road attribute data are fixed for each road segments for different time
periods and for weekday and weekend except the traffic volume and pedestrian data.
The current posted speed limits for site1, site 2, site 3 and site 4 are 40, 60, 80 and 80
km/hr, respectively.
Figure 1 shows the frequency distribution diagrams of observed speeds at site 1,
direction 1 for weekday at the time period 6:30-7:30 AM. Speeds higher than the
posted speed limit are regarded as “violations”. The cumulative percentage of viola-
tions (percentage of vehicles exceeding the speed limit) is illustrated in Figure 2.
100.00%
90.00%
80.00%
70.00%
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
6.30-7.30
9.30-11.30
15.30-17.30
18.30-20.30
22.00-23.30
Direction 1 Weekday
Direction 1 Weekend
Fig. 2.
Posted speed limit violations at site 1 direction 1
Development of FLM for 85
th
Percentile Speed Estimation
3
The development of the FLM starts with preparing the data sets for both weekday and
weekend. The road attribute data collected from the fields were used as the input va-
riables. The estimated 85th percentile of the operating speed was used as the output
variable. The input and output variables and their corresponding modified name used
in the FLM are shown in Table 1. Two separate models were developed (for weekday
and weekend). It is to be noted that volume to capacity ratio was also calculated from
the hourly traffic volume to incorporate in the FLM development.
The FLM development was done in two stages using the tool FuzzyTech [21] - first,
initial models were developed for both weekday and weekend by setting the member-
ships (fuzzy sets' parameters) and the knowledge base (rules) intuitively (using some
correlation analysis). To overcome the limitations of intuitive setting of knowledge base,
the neuro-fuzzy logic (integrated fuzzy and neural nets) was used [17].
