Using an Artificial Neural Network to Record and Analyse the Performance Indicators of a Transport Enterprise

Abstract

Recently, the issue of determining actual values of passenger flow processed by urban public and suburban transport on each section of the route in real time has become even more relevant since those values affect planning, organisation of operations and performance of transport enterprises.It is not possible to assess the real volume of passenger flow with the number of tickets issued to passengers due to the large number of unaccounted passengers (the discrepancy between the number of tickets issued to passengers and the real number of passengers transported).The objective of the study which results are described in the article was to develop an intelligent system for calculating and analysing the transport enterprise performance, which allows automatically collecting, processing and analysing information about passenger flows, calculating the passenger rotation ratio on the route, drawing up optimal bus schedules and timetables, adjusting vehicle traffic intervals, determining the need for rolling stock to minimise the likelihood that a passenger is denied boarding, improving the quality of passenger service and reducing the operating costs of a transport enterprise.Real-time calculation of passenger flow values for each vehicle on the route of urban public transport is carried out by a quantitative method based on artificial neural network technology, which allows automatic processing of a large amount of information from CCTV cameras installed in vehicle interiors.The use of theoretical research methods helped to create an intelligent system capable to analyse and compare the number of tickets issued to passengers with the actual number of passengers transported on Samara city public transport route, the results were displayed through a graphical user interface. It was possible to reveal dependences of the number of unaccounted passengers on the route on the amount of passenger flow and the time of day.A possibility of using the proposed intelligent system in commuter trains was noted, provided that the cars are equipped with video surveillance cameras.