Forecasting algorithm of time arrival with statistical data using based on the principle of templates selection
Artyom Olegovich Igumnov, Dmitriy M. Sonkin · 2016 13th International Scientific-Technical Conference on Actual Problems of Electronics Instrument Engineering (APEIE) · 2016
An approach to evaluation of en-route time from one point to another of a road graph using statistical data is described in this paper. Analysis of a vehicle dispatch monitoring system is performed. The task of forecasting of arrival time, its relevance and alternate solutions are described. The method of selection of historical data necessary for forecasting and the term “route template” are introduced. Template selection algorithm takes into account current road traffic conditions and uses the most suitable statistical data on route segment possibility for the moment of calculation. The fact that no temporary data is stored for each monitored vehicle as well as taking into account current traffic load conditions which makes it possible to minimize the loss of accuracy in case of unexpected traffic hindrance factors are specific features of the algorithm. Arrival time forecasting algorithm makes it possible to improve planning and scheduling for monitored vehicles delivering transport services inside the city. The algorithm offered was tested in the framework of the dispatch monitoring system of public municipal passenger transport of the city of Tomsk. Graphical analysis of comparison of forecast values obtained using the algorithm and actual arrival time is given in this paper, estimation error does not exceed 2 minutes.