Uncertainty Inclusive Runway Balancing Using Convolutional Neural Network

Abstract

This paper proposes a new optimization scheme using neural networks for runway balancing to minimize departure and arrival aircraft delay. The delay prediction for runway balancing optimization is obtained by a neural network, only without any additional simulations. Developing an accurate simulation model under an uncertain environment is difficult, but the proposed neural network model can estimate the average delay without modeling uncertainty explicitly. In this paper, the effectiveness of the proposed method is validated through numerical simulations. First, simulations are used to generate the data, which are then used to train the neural network. Next, the runway balancing problem is solved via simulated annealing using the delay predicted by the neural network. The simulation result shows that the proposed approach outperforms the simulation-based method under an uncertainty environment. Therefore, the neural network is shown to accurately estimate the delay under the uncertainty environment, which makes the proposed neural-network-based method applicable to objective function calculations for optimization.