Application of bi-level approach to traffic optimization.

The aim of this experiment was to apply bi-level optimization based on constraints coming from store-and-forward model and to test the resulting values for the duration of the green lights via a microsimulation. For the aim of this study two models were designed. The first model is based on bi-level optimization and the second model is the microsimulation in the traffic simulation software Aimsun. The results of the first model were calculated through a Python-based package named PAO and a solver. The optimized values for the duration of the green light were then entered in the Aimsun microsimulation model and tested with different traffic flows. The results show that the bi-level optimization may be reliable and fast model for a small network as the investigate one. However, additional testing with simulation software reveals some weaknesses as e.g. the bi-level optimization tasks lead to zero queues in front of traffic lights, but the software simulator shows approximately 20 vehicles mean queue for the network. [ABSTRACT FROM AUTHOR]