Recently, some studies have confirmed that the reproduction by simulation of user behaviour under different flow and geometric conditions, can identify a potential incident hazard and allow to take appropriate countermeasures at specific points of the road network [1], [2], [3]. In this paper a calibration and validation technique of a microsimulation model for short–term road safety analysis is presented. The microscopic model developed allows the estimation of road safety performance through a series of indicators (Crash Potential Index, Deceleration Rate to Avoid Crash, Maximum Available Deceleration Rate, Time to Collision, etc.), representing interactions in real time between different pairs of vehicles belonging to the traffic stream. When these indicators take a certain critical value, a possible accident scenario is identified. The calibration procedure was applied using an optimization algorithm to systematically modify the 5 parameters of the core behavior model (the General Motors car following model that is a module of the main traffic microsimulation model) in order to fit travel times obtained from simulations to the measured travel times. Experimental measures where obtained in one measurement site from a survey on a two lane undivided rural highway and were provided using a specifically developed video digital processing algorithm. In order to assess the capability of the microsimulation model to reflect reality and, consequently, to make further predictions, a validation technique was carried out. In the validation process the optimization algorithm shows the ability to increase the goodness of fit of estimated travel times to measured values. The estimated car following model parameters are used in estimating the goodness of fit with a set of observed data which have not previously been used in the calibration process. This procedure has brought very good results that show how the travel times estimation errors can be greatly reduced, even in different traffic conditions relatively to the calibration scenario, by using the set of parameters obtained in the optimization procedure.
Calibration of a new microsimulation package for the evaluation of traffic safety performances
ASTARITA, Vittorio;GUIDO, Giuseppe;Vitale A.
2012-01-01
Abstract
Recently, some studies have confirmed that the reproduction by simulation of user behaviour under different flow and geometric conditions, can identify a potential incident hazard and allow to take appropriate countermeasures at specific points of the road network [1], [2], [3]. In this paper a calibration and validation technique of a microsimulation model for short–term road safety analysis is presented. The microscopic model developed allows the estimation of road safety performance through a series of indicators (Crash Potential Index, Deceleration Rate to Avoid Crash, Maximum Available Deceleration Rate, Time to Collision, etc.), representing interactions in real time between different pairs of vehicles belonging to the traffic stream. When these indicators take a certain critical value, a possible accident scenario is identified. The calibration procedure was applied using an optimization algorithm to systematically modify the 5 parameters of the core behavior model (the General Motors car following model that is a module of the main traffic microsimulation model) in order to fit travel times obtained from simulations to the measured travel times. Experimental measures where obtained in one measurement site from a survey on a two lane undivided rural highway and were provided using a specifically developed video digital processing algorithm. In order to assess the capability of the microsimulation model to reflect reality and, consequently, to make further predictions, a validation technique was carried out. In the validation process the optimization algorithm shows the ability to increase the goodness of fit of estimated travel times to measured values. The estimated car following model parameters are used in estimating the goodness of fit with a set of observed data which have not previously been used in the calibration process. This procedure has brought very good results that show how the travel times estimation errors can be greatly reduced, even in different traffic conditions relatively to the calibration scenario, by using the set of parameters obtained in the optimization procedure.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
