Development of Online Vissim Traffic Microscopic Calibration Framework Using Artificial Intelligence for Cairo CBD Area

This paper makes a notable contribution to Transportation Planning in urban areas by considering the application of a key transportation Planning software package for the traffic conditions in the Central Business District of Cairo. For urban areas with heterogeneous and very congested traffic conditions and uneven driving behavior such development can be very useful. The paper shows how a microscopic simulation model using a Multilayer Feedforward neural network (MFNN) to calibrate online the VISSIM package driving parameters’ values based on predictions of travel time and traffic flow on the network elements. The two-step calibration procedure is faster and more applicable for on-line models than the approaches followed in current literature that require time-consuming iterations for model calibration. Also, this research uses combined Artificial Intelligence models (Long-Short Term Memory based Recurrent Neural Networks - LSTM-RNNs) and Multilayer Feedforward neural network (MFNNs) and calibrates them based on the driving behavior and traffic condition on successive time intervals. In this way, the prediction of the future traffic condition is based on actual traffic conditions on the past intervals and the actual driving behavior.