基于卷积神经网络的进近阶段飞机着陆距离预测OA

In order to prevent aircraft runway overruns during landing,a method of real-time landing distance prediction based on the fusion of convolutional neural network(CNN)and kinematic model is proposed.The landing characteristic parameters are extracted based on flight data during the aircraft approach phase,and the final approach speed is predicted by means of CNN model.The predicted speed is used as input to calculate the landing distance in conjunction with the kinematic model.Taking the Boeing 737-800 as an example,the predicted landing distance is compared with the actual landing distance and the influencing factors of the model are analyzed.The experimental results show that the prediction results are consistent with the actual situation,and the number of samples,time node to the runway threshold,and time window have significantly impact on the prediction accuracy.The larger the number of samples,the higher the prediction accuracy.When the sample size increases to 300,the prediction accuracy tends to stabilize.The closer the time node to the runway threshold,the higher the prediction accuracy.As the time window increases,the prediction accuracy shows a fluctuating trend,and the prediction accuracy reaches the maximum value when the time window is 10 s.This prediction method has high prediction accuracy and strong stability,with an average error of 4.54%,and can realize real-time prediction,which provides technical support for the research and development of the airborne landing distance early warning system.