Comparing and simulating travel time predictions from stationary and mobile sensors: Microscopic simulation of a stretch of the E4 motorway in Stockholm
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
This study looks at the two different traffic data sources available for traffic management: stationary sensors and probe vehicles. The strengths and weaknesses of each type of data collection system were briefly investigated in the literature review. A microscopic model using historical SRA (Swedish Road Administration) Motorway Control System data (MCS) from 23 March 2010 was created using a stretch of the E4 motorway in Stockholm as a study case. Two different traffic situations were examined: free-flow conditions between 12.00 and 13.00, and morning peak congestion, between 7.30 and 9.30. Models were calibrated by checking out the one-to-one relationship between observed and simulated values for two variables, average speed from each MCS detector station, and vehicle flows. MCS average speeds were calculated using the harmonic mean. The validation of the models was done using Theil’s inequality coefficient, as well as absolute and percentage errors. Two parameters that are inferred to have an effect on probe data quality, penetration and sampling rates, are shown to have an impact on the accuracy of travel time predictions. Probe vehicle fleets with a 60-second sampling rate and 1-percent penetration rate were shown to be adequate to retrieve traffic information from free-flow conditions. Probe fleets with a 5-second sampling rate and a penetration rate higher than 2% were shown to be sufficient to capture traffic information from congested conditions, and be competitive with the implementation of stationary sensors. Travel time predictions based on MCS data improve with increasing the number of detectors present in the system, for traffic going through morning peak congestion It is thereby concluded that probe vehicle data is an effective and cost-efficient method for collecting traffic data in both types of traffic conditions and can therefore provide useful input to traffic management systems.
Place, publisher, year, edition, pages
2011. , 65 p.
Engineering and Technology
IdentifiersURN: urn:nbn:se:kth:diva-40627OAI: oai:DiVA.org:kth-40627DiVA: diva2:441911