The Study Of Pilot Behavior Model For Analyzing The Safety Of Midair Encounter

In order to deal with the dramatic increase of air traffic,a new air transportation system called next generation air transportation system is planned to be built with improved capacity,safety and efficiency.For this purpose,dramatic changes are expected to be brought to current air transportation system.Thus,next generation air transportation system will be totally different in both system design and daily operation.Midair encounter may lead to midair collision which is the most fatal disaster in aviation.To ensure the safe operation of the new system,the safety of midair encounter should be analyzed thoroughly and accurately.Considering the high precision requirement and complexity of next generation air transporation system,advanced methods with improved efficiency are required to analyze the safety of midair encounter.This dissertation aims at developing advanced and efficient methods based on pilot behavior model to analyze midair encounter situation.First,Bayesian network is used to develop the model of midair encounter.Since human pilot is the key elelment in the model,a four stage(perception stage,cognition stage,decision making stage and response stage)pilot behavior model is proposed to model human pilot.Then,the decision making stage of human pilot is studied in detail.Based on the researches in the field of human behaivor,two decision making modes of human pilot in midair encounter situtaiton(analytical decision making mode and intuitive decision making mode)are achieved.This dissertation develops two decision making models for these two decision making modes respectively.In addition,in order to quantitatively analyze the effects of workload on midair encounter safety,this dissertation developes a workload predicting model based on queuing theory.This model is used to study the effects of pilot workload on midair encounter safety in various conditions.Finally,the proposed midair encounter model is extended to model multi-aircraft encounter in high density airspace,and the proposed pilot behavior model is applied to study aviation safety in high density airspace.The major work of this dessertation is summarized as follows:(1)This dissertation analyzes the components invlovled in midair encounter and their relations.Based on the analysis,a baysian network model is constructed to describe midair encounters.Through analyzing the researches in the field of human cognition,a four-stage(perception stage,cognition stage,decision making stage and response stage)pilot behavior model is proposed to describe human pilot which is the key component in the Bayesian network model.(2)Considering that deicision making is the most important process influencing midair encounter safety,two possible decision making modes(analytical decision making mode and intuitive decision making mode)of human pilots in midair encounter situation are derived based on cognitive continuum theory.The models for these two decision making modes are developed respectively: 1)the model for analytical decision making mode is developed using level-k relaxed strategy,and the computation complexity of this model is analyzed.The pilot behavior model with analytical decision mode is constructed through integrating the analytical decision model into four stage pilot behavior model.Based on this pilot behavior model,Monte Carlo simulation is used to demonstrate the application of this model in analyzing midair encounter safety.Finally,realistic flight data is used to examine the validadity of this decision making model;2)the model for intuitive decision mode is developed based on the concept of recognition primed decision making.The pilot behavior model with intuitive decision mode is constructed through intergrating intuitive decision making model into four stage pilot behavior model.Based on this pilot behavior model,Monte Carlo simulation is used to demonstrate how this pilot behavior model could be applied to analyze the safety of midair encounter.Finally,through comparing with the results of existing model and realistic flight data,the validity of this pilot behavior model is verified.(3)Considering workload is an important factor influencing the safety of midair encounter,a workload predicting model is developed leveraging queuing theory.Based on this model,Monte Carlo simulation is used to examine the effects of different human factors(rationality and compliance rate)on the safety of midair encounter and discuss the effects of decision supporting automation on workload.In addition,some suggestions on the introduction of decision supporting automation in next genenration air transportation system are provided.Finally,the validity of this workload predicting model is examined using the results of Monte Carlo simulation.(4)To analyze high density airspace in next generation air transportation system,the midair encounter model is extended to a Bayesian network model which is able to model large scale airspace with multi-aircaft encounter.Meanwhile,the quantitative performance metrics for this high density airspace are developed.Leveraging the Bayesian network model,the safety of high density airspace is studied.Especially,the effects of airspace density on the safety of high density airspace are investigated and the optimal choices of system parameters in high density airspace are discussed.