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.

This content is only available via PDF.
You do not currently have access to this content.