Control logic for the operation of rural variable speed limits subject to severe winter conditions

Variable Speed Limit (VSL) systems are an innovative way of managing traffic using Intelligent Transportation Systems (ITS). Unlike the traditional static speed limit signs, VSLs are speed limit signs with a capability of changing regulatory speed limits according to changing real time weather and traffic conditions. VSLs are primarily used to reduce traffic congestion and improve safety along roadways. Wyoming weather is highly variable and often severe, in the absence of recommended speed limits for such conditions drivers select their own safe driving speed leading to high variance among vehicles. This high speed variance, results in high crash rates and frequent road closures. Currently Wyoming Department of Transportation (WYDOT) using an interim manual protocol to deploy speed limits on VSL corridors. A new automated self-learning methodology for deploying VSL is described in this dissertation. As a first step in this project, weather variables that have a significant effect on vehicle speeds during bad weather conditions were identified using a robust regression statistical procedure. Second, an automated control strategy was designed that uses real time weather and observed speed variables. Third, a statistical model called decision trees was used in control strategy to enhance the performance of the automated control strategy. Fourth, a self-learning component is added to the control strategy to address concerns about the high variability of different storm events in this region. The proposed automated self-learning algorithm was tested on previous storm event datasets and results from the simulation indicate that the new system is more efficient than the current manual protocol of deploying speed limits. Finally, the transferability of the proposed control strategy was evaluated by applying the self-learning control strategy that was developed based on the storm data available at the Elk Mountain corridor to the storm data available at Laramie-Cheyenne corridor.