Predictive Probability Model of Pilot Error Based on CREAM

摘要

Prediction of pilot error is key of human-machine interface design in the cockpit, and is also an effective way on the reduction of accident ratio caused by human error. CREAM (Cognitive Reliability and Error Analysis Method) has been chosen to build the predictive probability model of pilot error based on investigation of various methods. The pilot error model built can be used not only to analysis the reason of accident but predict the error probability in particular scene. The model is validated through the experiment that pilots read the altitude during flight in different visibilities and time limits. The CPC (common performance conditions) including cockpit design, crew communication and other environment such as weather condition is always analyzed and calculated during the whole task analysis and then the reason of pilot error can be discovered qualitatively. The results are important for cockpit design to improve the airplane safety.