Model of Passenger Counting System Data Treatment

In recent decades, the technologies for automated monitoring passenger throughput have been developing at a rapid pace. One of the most popular types of equipment is presented by various detectors counting incoming/outgoing passengers allowing for assessing a rolling stock fill in real time. The archive data can be used to determine the passenger throughput and scope of work made. Therewith the input/output data allows calculating inter-stopping O-D matrix which, in its turn, allows assessing such characteristics as the average trip distance, turnaround of one passenger seat, etc. The analysis of measurement accuracy to estimate the passenger throughput on bus routes as well the data given by domestic and foreign producing companies showed that the maximum error (i.e., the difference between the aggregate number of incoming and outgoing passengers) could reach ± 15%. Therefore, to assess the inter-stopping O-D matrix the procedure resistant to spikes should be used. The paper addresses the methods that can be used to assess inter-stopping O-D matrix allowing for the increase of quality of processing the data supplied by I/O detectors.