The growth of the number of automobiles on the roads in China has put higher demands on the traffic control system that needs to efficiently reduce the level of congestion occurrence, which increases travel delay, fuel consumption, and air pollution. The traffic control system, urban real-time traffic control system based on multi-agent (MA-URTC) is presented in this thesis. According to the present situation and the traffic's future development in China, the researches on intelligent traffic control strategy and simulation based on agent lays a foundation for the realization of the system. The thesis is organized as follows: The first part focuses on the intersection' real-time signal control strategy. It contains the limitations of current traffic control systems, application of artificial intelligence in the research, how to bring the dynamic traffic flow forecast into effect by combining the neural network with the genetic arithmetic, and traffic signal real-time control strategy based on fuzzy control. The author uses sorne simple simulation results to testify its superiority. We adopt the latest agent technology in designing the logical structure of the MA-URTC system. By exchanging traffic flows information among the relative agents, MA-URTC provides a new concept in urban traffic control. With a global coordination and cooperation on autonomy-based view of the traffic in cities, MA-URTC anticipates the congestion and control traffic flows. It is designed to support the real-time dynamic selection of intelligent traffic control strategy and the real-time communication requirements, together with a sufficient level of fault-tolerance. Due to the complexity and levity of urban traffic, none strategy can be universally applicable. The agent can independently choose the best scheme according to the real-time situation. To develop an advanced traffic simulation system it can be helpful for us to find the best scheme and the best switch-point of different schemes. Thus we can better deal with the different real-time traffic situations. The second part discusses the architecture and function of the intelligent traffic control simulation based on agent. Meanwhile the author discusses the design model of the vehicle-agent, road agent in traffic network and the intersection-agent so that we can better simulate the real-time environment. The vehicle-agent carries out the intelligent simulation based on the characteristics of the drivers in the actual traffic condition to avoid the disadvantage of the traditional traffic simulation system, simple-functioned algorithm of the vehicles model and unfeasible forecasting hypothesis. It improves the practicability of the whole simulation system greatly. The road agent's significance lies in its guidance of the traffic participants. It avoids the urban traffic control that depends on only the traffic signal control at intersection. It gives the traffic participants the most comfortable and direct guidance in traveling. It can also make a real-time and dynamic adjustment on the urban traffic flow, thus greatly lighten the pressure of signal control in intersection area. To sorne extent, the road agent is equal to the pre-caution mechanism. In the future, the construction of urban roads tends to be more intelligent. Therefore, the research on road agent is very important. All kinds of agents in MA-URTC are interconnected through a computer network. In the end, the author discusses the direction of future research. As the whole system is a multi-agent system, the intersection, the road and the vehicle belongs to multi-agent system respectively. So the emphasis should be put on the structure design and communication of all kinds of traffic agents in the system. Meanwhile, as an open and flexible real-time traffic control system, it is also concerned with how to collaborate with other related systems effectively, how to conform the resources and how to make the traffic participants anywhere throughout the city be in the best traffic guidance at all times and places. To actualize the genuine ITS will be our final goal.
______________________________________________________________________________
MOTS-CLÉS DE L’AUTEUR : Artificial Intelligence, Computer simulation, Fuzzy control, Genetic Algorithm, Intelligent traffic control, ITS, Multi-agent, Neural Network, Real-time.
| Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:QMUQ.3525 |
| Date | 06 1900 |
| Creators | Wang, Lei |
| Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
| Detected Language | English |
| Type | Mémoire accepté, NonPeerReviewed |
| Format | application/pdf |
| Relation | http://www.archipel.uqam.ca/3525/ |
Page generated in 0.0021 seconds