机场跑道平整度, 评价指标, 飞行员评价, BP 神经网络

Based on the survey data of subjective evaluation of runway roughness by pilots on 37 actual test runways, conducted by the Federal Aviation Administration (FAA) in B737-800 and A330-200 flightsimulators, the relationship between the evaluation indicators of runway roughness and pilots′ evaluations of runway roughness in China was analyzed, and the impact of different aircraft models on pilots′ evaluations of runway roughness were compared and analyzed. Back propagation (BP) neural network was built, the current runway roughness evaluation indicators in China and aircraft gross weight (AGW) were taken as the input, and the pilots′ acceptance of runway roughness were

taken as the output to predict the pilots′ evaluation of runway roughness. The results showed that the goodness of fit

between each runway roughness evaluation indicator and the pilot′s evaluation of runway roughness is low, making

it impossible to predict the pilots′ evaluation results separately. The aircraft type can affect the pilots′ evaluation of

runway roughness, and the characteristics of the aircraft type should be considered when the pilot evaluates and

predicts runway roughness. The BP neural network has a prediction accuracy of 100% in the training set and 95.5%

in the test set. It can effectively integrate the characteristics of China′s runway roughness evaluation indicators and

achieve accurate prediction of pilots′ runway roughness evaluation results across aircraft types.

airport runway roughness, evalution indicator, pilots′ evalution, BP neural network