INCIDENT DETECTION

Persaud ID Algorithm

Background

The basic concepts behind the Persaud algorithm for incident detection can be summarized as follows: based on the analysis of traffic data collected at each detector station for traffic conditions with and without incidents, the fundamental volume-occupancy space of the traffic diagram for the corresponding detector station is split into several regions which correspond to the various possible traffic states, i.e. uncongested, congested due to recurrent congestion, congested due to an incident, etc.

There are several parameters and functions that define the boundaries between the regions in which the space is partitioned, some of them with a theoretical background supporting their definitions such as:

• OCRIT: the maximum occupancy for the observed uncongested flows;
• VCRIT the maximum uncongested flow;
• LUD the lower bound on uncongested data (i.e a logarithmic function of the form: k*a*OCC^b)

While some others like the so-called OCRIT-10 and the factor k included in LUD are defined only on empirical grounds after a trial and error process. When an incident occurs in the space between two detectors, the Persaud logic combines the traffic states of the downstream and upstream detectors in an attempt to identify congestion (analysis of traffic states at upstream detector) and once congestion is identified discriminate the recurrent congestion from the congestion caused by an incident analysing the states of the downstream detector.

The Persaud ID algorithm was developed by UPC.