Research On Dynamic Traffic Characteristics And Operation Situation Of Terminal Area

According to the rapid increasment of air transportation demand,using technological means to promote ATC operation efficiency and service ability is an effective way to redress the imbalance between the supply and demand at present.The traditional management mode based on experience is relatively extensive.Air traffic characteristics,regularity,temporal and spatial distribution characteristics are not effectively used to help understanding operation problems,so it’s difficult to achieve refined management and scientific decision-making.With the constant improvement of ATC operation data acquisition,the basic data condition has been established for implicit knowledge extraction by data mining technology.The implicit knowledge includes dynamic rules,cluster patterns scene classification and so on.Intelligent data analysis and decision-support technology is becoming research hotspot and trend in the field of air traffic.This paper comprehensively introduces current research situation and trend of traffic characteristic analysis and intelligent descision-supporting technology in the field of air traffic.Based on different ATFM decision-making demand,several key methods or models,which focus on traffic characteristic analysis and situation identification,were developed by using data mining technology.The research contents and achievements mainly include:(1)The distribution characteristic and changing rules of taxiing phase operation time were ayalyzed,and a measurement of bad weather impact on the departure flight was established.According to the different phases of flight,firstly,a clustering algorithm was applied to analyze the clustering pattern,distribution characteristics of operation time in airport airside taxiing phase,and probability statistics method and concept of “anomaly” were used to extract the periodic variation rules of airborne flight time.And then,an estimate method was established to measure operation time of flight legs,with creative steps of calculating individual segment separately and then integrating them accordingly,which could support the flight plan scheduling in strategic and pre-tactic phases.Further,in order to pedict the delay of departure phase impaced by severe weather at airport,an improved Weather Impacted Traffic Index(WITI)suit for airport was designed and used to analyze the relationship with different departure delay indicators.At last,a set of regression models were builded,which could support the early prediction of departure delay based on expected meteorological conditions(2)A terminal traffic flow identification method and a traffic flow phase-state discriminant model were established.According to the demand of air traffic flow’s characteristics analysis and phase-state identification,firstly,an identification method for different kinds traffic flow in terminal area was proposed based on trajectory similarity model design and spectral clustering.Reference trajectory for each traffic flow was also extracted by kernel density estimation.Then,a set of characteristics were defined to describe taffic flow phase-state.The distribution relationship and change rules between characteristics were revealed to support the recognition of 3 phase-states of traffic flow: free,steady and congestion,operation feature of different phase-state and transition law were extracted as well.Finally,a fusion method based on factor analysis and genetic EM clustering was developed to analyze the main influences and extract recessive characteristic vector of traffic flow,ultimately help to identify phase-state of traffic flow.Further,it also provides a basis for arrival/departure procedure design and traffic flow allocation in terminal area.(3)A fuzzy evaluation method and a situation prediction model for terminal area traffic situation were proposed.In order to slove the problem of indicator selection and subjective information in traditional methods,a set of macroscopic and microscopic traffic characteristics were extracted to build the description of traffic situation from 3 dimensions.Then,a fuzzy evaluation method based on measure of medium truth degree and entropy theory was introduced to calculate and classify the traffic situation of terminal area objectively.Futher,consider the fuzziness and cognitive differences of traffic situation,and the demand of traffic situation prediction in real-time AFTM,a back-propagation artificial neural network(ANN)model was desgined,trained and validated by abundant typical samples.The result verified the effectiveness of ANN model on rapid prediction of traffic situation.(4)A conceptual framework and scalable system architecture for ATFM descsion-support of terminal area were designed.Firstly,base on the target and contents of ATFM,the application scope of descsion-support technique and corresponding method system were defined.Then application scope and specific method associate with different ATFM phases,and the conceptual framework was set up across two dimensions: application scope and ATFM phases.Further,according to the general data mining process,a hierarchic data-extraction model was established,and a multi-layer architecture characterized by scalable and loose coupling was designed,which solve the separation among the descsion-making logic,data-mining logic and characteristic calculation logic.So the architecture could support rapid application development and configuration.At the end of the paper,the innovations and main research results were summarized.Base on the research limitions and application demand,future research content and direction were prospected as well.