Research On Pilot Modeling For Flight Safety Quantification

Improving flight safety has been an important topic. Statistics fromdifferent sources indicate that about60%-80%of the total flight accidentsare caused by human errors. It is significant to study the relationshipbetween the human factors and flight safety. The paper is supported by theNational Key Basic Research and Development Program of Support(NO.2010CB734103).The most effective way to study the quantitative impact of humanfactors on flight safety is building an integratedHuman-Machine-Environment simulation model for investigating therelationship between human factors and flight safety. The key technicaldifficulties lie in constructing a model that contains the characteristics ofhuman such as the adaptive behaviors, uncertainties and the constraints ofmultiple human factors.The paper mainly include the following aspects: analyzing the humanand machine function assignment, analyzing the structure andshortcomings of the pilot model exists; building a six degrees of freedomnonlinear aircraft model; designing a longitudinal and transverse planesheading controller; proposing a new pilot model that includes threeinformation processing modules: Information Perception, Decision Makingand Action Execution. By selecting different probability distributionfunctions for the human factors and selecting several different control rules,the pilot model can reflect people’s different control tendency; investigating the impacts of uncertain human factors on flight safetythrough constructing two different scenarios.The results of computer simulations show that all these selectedhuman factors have impacts of flight safety and the effects can bequantitatively analyzed. In particular, a large Indicator-Scanning-Sequencewill have negative influence and the impacts of Motor-Error andIndicator-Scanning-Sequence are greatly different in contrast with thebaseline experiment; Simulation-based analysis shows that flightperformance can be dramatically improved with the quantitative humanfactors which are quantified via boundary identification of flightperformance margin.