Model Based Fault Detection of Freeway Traffic Sensors

Abstract

This paper presents a model based fault detection and exclusion scheme that implements a decision logic to automatically identify faulty or mislocated freeway traffic sensors in the presence of unknown on-ramp and off-ramp flows. The algorithm is deployed within the framework of a suite of software tools, named TOPl, which models traffic flow via a macroscopic model, calibrates the model based on available data and runs simulations to evaluate various operational strategies such as ramp metering, demand management, incident management, etc. TOPl has been used to model various freeways in California, such as Interstate 80, Interstate 210, Interstate 880 and Interstate 680. Two main difficulties with data collection on California freeways were found to be missing ramp flow and faulty mainline data, which decrease the accuracy of the model and increase the time and effort invested in model calibration. The former of these difficulties has been previously addressed by an iterative learning algorithm that estimates the missing ramp flows and the latter is tackled by the method presented in this work.

On the Fundamental Diagram for Freeway Traffic: Exploring the Lower Bound of the Fitting Error and Correcting the Generalized Linear Regression Models

“…Greenshields [1] first proposed a linear model to describe the relationship between speed and density and made a pioneering work in this field. This rudimentary relationship has since been refined through the introduction of numerous models [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18]. These studies try to define precise relationships, utilizing practical parameters to reflect the traffic flow features more accurately.…”

On the stochastic fundamental diagram for freeway traffic: Model development, analytical properties, validation, and extensive applications

“…As flow is the product of speed and density, this relationship usually refers to flowdensity or speed -density relationship. Since the seminal Greenshields model (Greenshields et al, 1935) was proposed, numerous studies have been done to improve this over-simplified relationship empirically and/or analytically (Greenberg, 1959;Newell, 1961;Underwood, 1961;Edie, 1961;Kerner and Konhä user, 1994;Del Castillo and Bení tez, 1995a&b;Li and Zhang, 2001;Wu, 2002;MacNicolas, 2008, Ji et al, 2010Wang et al, 2011;Wu et al, 2011;Dervisoglu, 2012;Keyvan-Ekbatani et al, 2012&2013). The main focus of these studies is to develop accurate deterministic speed-density models with two or three practically meaningful parameters 1 .…”