Naslov
Performance Evaluation of Advanced Algorithms for Traffic Flow Forecasting

Autori
Krljan, Tomislav ; Grbčić, Ana ; Poletan Jugović, Tanja ; Lopac, Nikola

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
International Conference on Sustainable Transport | Book of Abstracts / Vukelić, Goran ; Brčić, David - Rijeka : Pomorski fakultet Sveučilišta u Rijeci, 2022, 58-58

ISBN
978-953-165-138-7

Skup
International Conference on Sustainable Transport (SuTra 2022)

Mjesto i datum
Opatija, Hrvatska, 29.09.2022. - 01.10.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
traffic flow forecasting ; traffic management ; decision support systems ; time-series forecasting algorithms

Sažetak
In moderately developed cities, the realization of mobility desires today still depends heavily on the private transport system. Transitional cities, where demand for private transport modes is currently highest, face a number of problems. These problems are typically related to the level of traffic congestion, which leads to unacceptable travel times, fuel waste, lower operational efficiency, poor air quality, high driver stress levels, and safety concerns. Therefore, the need for effective short- and medium-term traffic flow forecasting is becoming increasingly important for the development of decision-support systems and the implementation of low-error operational procedures in traffic management. This paper evaluates the performance of advanced time series forecasting algorithms that ensure accurate future input data for traffic management decision making. The algorithms were trained, validated, and tested on a dataset of hourly traffic flow on a divided multilane highway (urban bypass) in 2021 (Rijeka, Croatia). The dataset consists of traffic flow data collected from inductive loop detectors on 10 highway sections (two detectors per driving direction ; one detector on the right driving lane and one detector on the overtaking lane). The accuracy of the evaluated algorithms is assessed using performance metrics such as Mean Square Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). All tested algorithms meet the application criteria for detecting diurnal trends in traffic flow and are suitable for implementation in traffic management decisions.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo, Tehnologija prometa i transport

POVEZANOST RADA