Research On Complex Network And Network Dynamics For Air Tranportation Systems

From nature to society, many real systems can be described as networks. Complex network theory and network dynamics theory provide a suitable tool to investigate the network topology and dynamical processes, including the air transportation system and airway traffic process.Since the first scheduled services between Tampa and St. Petersburg in year 1914, the airway traffic has been favored due to its remarkable speed advantage. Nowadays, the"fast","safe","flexible"airway traffic plays a more important role in national economy for both developed and developing countries, especially in the stage of emergency-rescue (earth quake, tsunami, volcanic eruption etc.).The air transportation system can be naturally denoted by networks. In the macro-level, the airports can be denoted by vertice and the airlines between airports can be denoted by edges; in the micro-level, the aircraft can be denoted by vertice and the detection between aircraft can be denoted by edges. With the help of complex network theory and network dynamic theory, we can investigate the structure and dynamical properties of air transportation system, alleviate the air traffic congestion, and finally improve the efficiency and safety of air flights.The airline network of China (ANC) has developed from star-like topology to hub-radiate-type topology with four central hubs (Beijing, Shanghai, Shenzhen and Guangzhou). Based on the complex network theory, the topology of ANC is systematically investigated. It is found that ANC is a small-word network with two-regime power-law degree distribution and the traffic flow on ANC rapidly increases with an exponential function. The increment of air traffic flow will surely cause the air traffic congestion, which will induce the low efficiency of air traffic and raise the risk of aircarft conflict. Based on the network traffic dynamics theory, we prposed a dynamical delivering capability mechanism and a queuing rule based on remaining pathlength to improve the network traffic efficiency. It is found that the network capcacity is remarkably enlarged and the probability of aircarft conflict is greatly reduced.The macro-level regulation (such as the topopogy adjustment of ANC or the air traffic management) can balance the air traffic flow, but the aircraft conflict is still inevitable, sometimes very frequent. How to solve the aircraft conflict becomes a hot topic in recent years. In this thesis, the aircraft conflict resolution problem is modeled in the framework of evolutionary networked game theory, and three key factors in the model are deeply investigated:( a ) Network Structure : The weighting parameter is introduced into evolutionarygame model to denote the heterogeneity of aircraft's detection region. It is found that the system behavior is remarkably influcenced by the weight of aircraft: the heterogeneous of detection region can induce heterogeneous degree distribution, the heterogeneous degree distribution can induce high cooperation level, and high cooperation level means less aircraft conflicts. Besides, the hub aircraft with high degrees play an important role in the evolution. Controlling a few hub aircraft's behavior can greatly improve the efficiency of conflict resolution.(b)Game Model:The standard game model is modified to represent the heterogeneity of aircraft: if the adjusting parameter is positive, the preponderant aircraft can dodge conflict with a small deflection angle while the disadvantaged aircraft have to dodge with a larger angle. Although the positive parameter increases the cost of many disadvantaged aircraft, the system efficiency is improved due to the network-feedback mechanism. A new fitness evaluation method is proposed. It is found that the cooperation level can be promoted when the surrounding factor is considered.(c)Learning Rule:The strategy learing rule plays a crucial role during the evolution process. The strategy learning process can be devided into two steps: neighbor selection and strategy updating. In the neighbor selection step, a non-linear preferential neighbor selection rule is proposed and it is found the system efficiency is improved if the aircraft are more likely to select the preponderant aircraft. In the strategy updating step, a history-based strategy updating rule is proposed and it is found the consideration of history memory can help the aircrafts make better choice to avoid air conflict more efficiently.The investigation of present thesis is a cross subject of aviation, computer science and statistical physics. Based on the complex network theory and network dynamic theory, the air traffic congestion problem and aircraft conflict resolution problem are thoroughly studied. The results can deepen our understanding the evolution discipline and traffic behavior of air transportation system, and provides a novel way to solve the aircraft conflict problem.