Research On The Complexity Of The Traffic Behaviors In Air Traffic Management

Air traffic system is both the object and the product of ATM (Air Traffic Management), and itsattributes form therof and their emergence could be explicitly reflected by the observeable air trafficbehaviors. Therefore, it’s much important to analyze traffic behaviors with their internal mechanismsas a whole for getting knowledge of modern ATM syetem and objectively assessing its performance.As an initiative discussion, the paper proposed a methord framework studying traffic behaviors withtheir complexity, which consisted of concept system refining, metric system establishing and metricsystem analyzing, to reveal the mechanisms of traffic behaviors and set up fundamentals developingautomatic ATM system from different decision-making level. The paper firstly discussed the conceptof traffic behaviors and refined meanings of their complexity by introducing complexity scienceepistemology into analysis of formation mechanisms of behaviors such as en-route queue, conflictavoidance and asymmetry situation information distribution, which clarified the objects categories inthe research field of ATM complexity. Then, a complexity metric system of traffic behaviors wasinitiatively established, restructured and refined to be useful tools analyzing traffic behaviors and theircomplexity. Last but not least, an analysis method system towards at static and dynamic trafficbehaviors with their complexity was proposed using the complexity metric system under general andpreferred decision-making modes from multi-dimensional points of view. The methord frameworkproposed above is helful to refine and explore the operational characteristics and their modes of airtraffic system, so as to concrete the fundamentals of automatic ATM system development satisfyingdifferent specification requirements. Detailly, major subjects and corresponding results involved in thepaper were outlined as below:(1) A unidirectional route queue model was proposed to discuss formation mechanisms of suchtraffic behaviors that a route was regarded as a single service station under definitely simple flightrules and randomness. Then, a multi-agent conflict resolution algorithm was proposed to discussemergent traffic behaviors introducing two dimensional flight freedoms under multi-aircraftsnon-cooperative strategic distributed conflict resolution mode. A combined maneuvering conflictresolution strategy model was further proposed to discuss emergent traffic behaviors as large amountsof aircrafts cooperatively strategicly resolve conflicts while pursueing distributed decision-making.Afterwards, an asymmetry information distribution model was proposed to discuss the effects ofhuman factor on inducing the asymmetry distribution situation of the traffic information. All themodels and algorithm investigated indicate nonlinear effects of the factors in the traffic system are immanent, and simplely FCFS flight rules, multi-dimensional flight freedom, high density traffic flowand decision-making priority closely interact to emerge regular but self-contradictory traffic behaviors.As a result, four aspects concerning implications in the traffic behaviors concepts, especially as fortheir complexity from human factor and automatic machine points of view, were abstracted andoutlined, which provided the explicit category of the object in the research field of ATM complexity.(2) A complexity metric system towards at traffic behaviors was proposed consisting of metricshelpful to quantitatively depict operational status of air traffic system and its evolution, consideringhuman being and automatic machine factors from multi-dimensions as complexity scienceepistemology. The metrics were either selected directly from literatures, or produced by abstractingreasonable parts of existed metrics for performance evaluation. The system could be classified intoairspace and traffic sub-systems., and the later sub-system included density, dynamic and conflictmetrics. Cases analyses indicate greater number of traffic metrics is time related and fluctuate inlayers, and the range differences among metrics are large. Though the metric system is helpful toexam traffic behaviors locally to describe the operational situations of air traffic system fromdifferent points of view and levels, it couldn’t comprehensively analyze air traffic system with acomplete explaination framework.(3) A metric system restructured method was proposed to unify the expression of each metric,from which a metric system refined method was further proposed abstracting the characteristic vectorto hierarchically present and utilize the metric system to make it an effective tool for traffic behaviorsanalysis with their complexity. Object framework model was set up to provide a unified data structurefor different metrics with complete traffic information presented by one image sketching relationshipseither between a metric and its base object, or among different base objects. The characteristic vectorof an object framework model was abstracted based on its moment function to achieve hierarchicalrefinement of the metric system with less resource requirements for data processing and storage andinformation loss and distortion. Cases analyses indicate restructured metric system not only provide aunified analysis reference for the metric system, but also present traffic behaviors in holographic anddirect viewing manner. The characteristic vector of an object framework model could degrade at leastone order of magnitude in amount of data storage so as to decrease input datas directly for trafficanalysis and pattern recognition.(4) An analysis method system based on the metric system restructured and refined method wasproposed to systematicly analyze metric calculation results and take advantage of automatic ATMsystem processing and managing large amounts of traffic data and exploring special modes of airtraffic behaviors. A non-supervised division cluster analysis method was initially proposed to dividethe refined metric system into several subsets automatically, and static traffic behaviors in a time period were holographicly sketched by cluster varibles such as “subset number”,“difference betweensubsets”,“difference in a subset” and “subset containment”. After a difference function was defined totransform the cluster varibles “subset containment” into a new one varible “difference betweensubsets containment”, it was used together with “subset number” and “difference between subsets” toreveal the order of dynamic traffic behaviors by investigating dynamics similarity of three variables’synchronicity defined as “synchronicity complexity” under information meanings. Cases analysesindicate “subset number”,“difference between subsets” and “difference in a subset” could effectivelydescribe relationship intensities of different static traffic behaviors, while “subset containment” couldprovide details of specific relationships of different static traffic behaviors.“Synchronicitycomplexity” could provide decision makers effective measure tool to analyze synchronicityrelationships of different characteristics of dynamic traffic behaviors at more suitable scales.(5) A preference function was defined after revealing the internal mechanisms of preferreddecision making, and used to analyze static and dynamic traffic behaviors with their complexity underpreferred decision-making modes based on the restructured and refined metric system. Afterwards, acorrelation information entropy model was set up to make comparision of traffic behaviors modes inthe objective and subjective air traffic system, which complete the analysis method system based onthe hierarchical and multi-dimensional complexity metric system. Cases analyses indicate automaticATM system could independently process traffic data for pattern recognition because preferreddecision-making makes little effects on revealing synchronicity relationships of differentcharacteristics of dynamic traffic behaviors from information aspect. And compared to sector13,sector5possessed of lower synchronicity complexity in Guangzhou air traffic system in the case day.