Model-Based Traffic Control with Different Traffic State Measurements

In the research described in this paper, traffic flow is controlled using dynamic speed limits, obtained by Model Predictive Control (MPC). MPC is a model-based approach, where the states of the system, influenced by control actions, are predicted over a certain time span. The states of the system are the mean speeds and densities on the motorway. Traffic flow models typically use space mean speeds, while measurements on motorways are often obtained by loop detectors, returning time mean speeds. Several methods for obtaining estimates of the space mean speed based on measurements on a fixed location are discussed, and the possible performance loss of using another mean speed than the space mean speed for model-based traffic control, is investigated.

The objective for the traffic controller is minimizing the Total Time Spent (TTS) by vehicles on the motorway. When shock waves exist in a traffic network, travel times are increased compared to the free flow situation, since vehicles have to slow down. Using dynamic speed limits, shock waves can be reduces, or even dissolved. The simulations on which we report in this paper show an improvement in the TTS of 15.6%, thereby increasing the flow by nearly 5%.