Authors:
Lan Ma
;
Weian Li
and
Zengxian Geng
Affiliation:
Air Traffic Management College, Civil Aviation University of China, Tianjin and China
Keyword(s):
Air Traffic Safety, Rough Set, Attribute Reduction, BP Neural Network.
Abstract:
The safety of air traffic control is an important link in the safety system of civil aviation industry. In order to evaluate the safety risk of air traffic control in a more comprehensive and reliable way, proposing an air traffic safety risk modeling and evaluation method based on rough set and BP neural network. After analyzing the factors that may affect the safety in the actual work of ATC, 24 attribute variables which can measure the safety risk of ATC are given. Aiming at the shortcomings of traditional neural network training with high redundancy, slow convergence and easy to fall into local optimum, the attribute reduction method is used to reduce the input attribute by rough set theory. Under the premise of not affecting the training results and the accuracy of the data, removing the low correlation attributes with the results, the network structure is simplified, the training times are reduced, and the training speed and accuracy of the neural network are improved. Use the
simplified condition attributes of the original data after rough attribute reduction as input data, the conflict resolution object is as output data, using MATLAB to build the neural network, and the trained network is tested and verified to be reliable. Compared with the model before the reduction of the initial data, significantly improves the accuracy and efficiency. The model is verified by examples The results show that the combination of rough set and BP neural network can accurately evaluate the risk of air traffic control, change the risk assessment from qualitative to quantitative, and provide guidance for the actual operation.
(More)