Abstract
Microscopic traffic simulators, such as VISSIM, use time-based step-by-step vehicle movements to predict outcome of proposed network configuration. This process is inherently slow and impractical for use in stochastic large network optimizations. Our solution is to use a mesoscopic traffic simulator with one-step movement approach between nodes in traffic network. Our model uses a predictor function for each road segment which predicts travel time distribution for traffic conditions using machine learning (AI). Our simulator chooses a random sample from predicted distribution as travel time. Travel time distributions may be sensitively dependent on parameters such as traffic and road conditions, and traffic behavior patterns which may be dependent on specific road. Experimental results show that our simulator's fidelity is like that of VISSIM for various traffic conditions. We demonstrate that our simulator is more than 100 times faster than VISSIM and provides network performance results that are comparable to these models.