Forecasting flight time based on BP neural network

An accurate estimated flight time is essential to modern air traffic management systems. Because the forecast is associated with many factors and needs large numbers of statistical calculation, the traditional methods used to forecast flight time are limited and inadequate. In this article, a back propagation neural network model is presented for forecasting the flight time. Firstly, the main factors impacted on flight time were analyzed and the air traffic control and weather condition factors are input to the model as the key factors. Then the optimal number of hidden nodes was obtained by Bayesian information criterion for speeding up the convergence of BP networks. Simulation results show that the method has rapid convergence and good scalability to accurately forecast flight time.