Optimal signal control with multiple objectives in traffic mobility and environmental impacts

• Main Author: Robles, Danny
• Format: Others
• Language: English
• Published: KTH, Trafik och logistik 2012
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-101927

The increasing number of motor vehicles in urban areas worldwide requires a smart traffic management establishing sustainability on the traffic system. Traffic signal control is a powerful tool in this field since it can control flow patterns in urban areas. Historically, traffic signal optimization was applied to satisfy the goals in mobility of traffic systems e.g. measured by travel delay, stops etc., and very little is known if such a strategy would be optimal for system sustainability in terms of emission and fuel usage. The thesis focus on finding the trade-offs between mobility and impact measures and compares these with approximated real signal strategies. The research objective of the thesis is to create a multi-objective computa- tional framework based on the integration of a microscopic traffic simulation model with a micro scale fuel and emission model. The proposed framework is able to implement mobility and impact objectives in a multi-objective opti- mization process. The microscopic traffic model VISSIM is used to simulate the traffic and two different emission models, CMEM and VT-Micro, are used to estimate the vehicular emissions and fuel consumption. The optimization is based on NSGA ii, a multi-objective genetic algorithm. The proposed framework is demonstrated by conducting two case studies, a single intersection in Wuhan and two coordinated intersections in Stockholm. The investigated objectives used in the optimizations are network delay, av- erage number of stops and average fuel consumption. Moreover, the best solution of each objective is subjected to a emission evaluation. Due to time consuming optimization processes, an upper limit of iterations is set for both cases. All simulations are based on 60 minutes of traffic simulations with additional 15 minutes for warm up. The study shows that the proposed framework is successful in finding signal control strategies producing better values of the investigated objectives com- pared to the real signal approximations. One could also see apparent trade-off between mobility and sustainability depending on the selected objectives.