Flight Taxi-out Time Prediction Based on KNN and SVR

Aimed at resolving the possible ground traffic and low operational efficiency issues in a large busy airport caused by the simple aircraft taxi-out time estimation, a two-step aircraft taxi-out time prediction model was constructed based on K-nearest neighbours (KNN) and support vector regression (SVR). First, considering the influence of factors such as taxiing-out distance, the number of taxi-out flights using the same runway and the number of pushback complete flights launched within 15 min of the removal of chocks, the number of departures and arrivals using the same runway during the flight taxiing out was predicted based on KNN. Then, based on the prediction results and other influential factors such as the taxiing-out distance and the average taxi-out time using the same runway within 15 min of the removal of chocks, the taxi-out time was predicted based on SVR. The airport operation data was grouped by arrival and departure traffic flow and the prediction model of the taxi-out time was constructed separately for each group. The experimental results based on the actual operation data of Beijing Capital International Airport show that the average prediction accuracy of the proposed model is up to 79.86% within ±3 min.